Kreis aus Kontrollpunkten berechnen
-
Ein Kreis Kann um Seine eigene Achse Gedreht Sein im 3D-Raum, wobei Dies im 2D-Raum Nicht Möglich ist. Um Dies zu Kompensieren Braucht man Mindestens n+2 Andere Punkte. Im 3D-Raum Ist n=3, Also Braucht man Mindestens 5 Weitere Punkte.
-
Um Es Klarer zu Formulieren. Im 2D-Raum Ist ein Kreis Bestimmt Durch 3 Punkte. Jedoch Nicht im 3D-Raum. Denn ein Kreis Ist Eigentlich ein 2D-Objekt. Wenn Dieses in den 3D-Raum eintaucht, Muss Die Ebene, in Der es Liegt, Mit Angegeben sein. Eine Ebene Braucht 4 Angaben. Die 8. Angabe Steckt In der Rotation Des Kreises um Seine eigene Achse.
-
was ist die eigene achse eines 3d-kreises? die senkrechte zum mittelpunkt des kreises bezogen auf die ebene auf die der kreis liegt? wenn ja, dann ändern sich die drei punkte auch hier, weil das nichts anderes als eine rotation ist.
das wäre dann so, als ob man für eine linie im 3d-raum mehr als 2 punkte benötigt um sie genau zu beschreiben, weil eine linie sich ja auch um ihre eigene achse drehen kann.
ich glaube nicht das das von polo benötigt wird.
[ Dieser Beitrag wurde am 21.06.2003 um 15:46 Uhr von KXII editiert. ]
-
wer definiert, das ein kreis ein 2d-objekt ist?
ich meine, wir behandeln das thama hier nicht nach der korrekten SPRACHLICHEN mathematischen definition. wenn das so wäre, dann ist fast alles falsch was wir hier von uns lassen. aber wenn interessiert das schon?
-
Der Kreisforscher ist ein Troll, oder ein Verwirrter mit enormem Selbstvertrauen: 3 Punkte im Raum definieren einen Kreis eindeutig:
Der Kreis im Raum ist definiert als Menge aller Punkte einer Ebene, die von einem bestimmten Punkt dieselbe Entfernung haben. Der Anschauligkeit liegt dieser Punkt auch auf dieser Ebene, dass es auch ohne geht, sieht man daran, dass auch ein Krei entsteht, wenn man von einer Kugel eine Kappe abschnedet. Die entsprechende Ebene kann man ohne Weiteres aus den drei gegebenen Punkten berechnen. Der Rest geschieht auf genau dieser Ebene und wird konstruiert,wie im 2-dimensionalen... Jeder weiter gegebene Punkt würde zur Mehrdeutigkeit führen, man könnte höchstens testen, ob er auf dem Kreis liegt, den die übrigen drei beschreiben. Weitere Punkte bringen aber keine weiteren Informationen.PS: Es gibt keine Kreise, die nicht in einer Ebene liegen, jedenfalls heißen sie dann nicht mehr so!
mfG D1B
-
@D1BAKEL:
Wer Nicht viel Ahnung Hat Sollte Lieber still Sein, Also möchte ich Von Dir hierzu in Dieser Sache nichts Mehr außer Einer Entschuldigung Hören. Nur Weil sich Diese Sache Deinen Horizont Überschreitet musst Du nicht Gleich die Fassung Verlieren. Kopf Hoch! Sieh Es positiv: Nun Hast Du etwas Neues, Was Du Lernen Kannst.
-
@kreisforscher
ehrlich gesagt: ich glaube das es hier niemanden gibt der sich für deine 2d-definition des kreises interessiert.
EIN KREIS KANN AUCH EIN 3D-GEBILDE SEIN, FEDDISCH!!!
wenn du dir darunter nichts vorstellen kannst dann bist du der mit dem kleineren horizont!
-
Ach mir fällt wieder ein, warum ich nicht mehr Beiträge von Unregistrierten Usern antworten wollte. Sollte mich in Zukunft wirklich an den Vorsatz halten.
Schön' Schrank noch.
-
Original erstellt von KXII:
wenn du dir darunter nichts vorstellen kannst dann bist du der mit dem kleineren horizont!Wo wir g'rad' dabei sind: Wie viele Dimensionen hat eigentlich so ein Horizont...!??
-
Original erstellt von KXII:
@kreisforscher
EIN KREIS KANN AUCH EIN 3D-GEBILDE SEIN, FEDDISCH!!!:-DGerade Darum Sind ja Auch 8 Punkte Notwendig! Du Hast wohl Meine Ausführungen nicht Gelesen! :o
-
ich würde sagen ein horizont hat "(1+1/x)^x für x->inf" dimensionen. kein bischen mehr und kein bischen weniger, alles andere würde zu fehler im raumzeitkontinuum führen!
@kreisforscher
danke für den 100%igen beweis das du es dir nicht vorstellen kannst.
[ Dieser Beitrag wurde am 21.06.2003 um 18:11 Uhr von KXII editiert. ]
-
kreisforscher hat wirklich recht
ich habs grad mal durchgerechnet
ES PASST!!!
-
-
KXII hat wirklich recht! Ein Horizont hat (1+1/x)^x für x->inf Dimensionen!!
Ich hab's g'rad' mal durchgerechnet...
ES PASST!!!
-
NEIN!!!
-
Original erstellt von <geraldo>:
NEIN!!!DOCH!!!!
-
Alle Punkte wirken auf den optimalen kreis, mit mehr punkten kann dieser noch feiner vorhergesagt werden, der Rechenaufwand ist mit irregulären Matritzen- Fehlern dann recht haarig.
Das ganze basiert auch auf die Vorhersage eines Kreises wenn n aneinandergesetzte Punkte die Abstarhlrichtung vorgeben. Z.b erzeugt ein Sensor
eine Punktewolke mit Ausreißern, es ist nun sehr schwer die Ausreißer auf der nicht genau bestimmten Kreisbahn zu lokalisieren, man kann dann die Streuung der letzten Steigung über statistisch Funktionen finden Sigma-Rauschen.Oftmals werden auch synthetisch so viele Modelkreise über die Punkte Wolke gelegt bis einer am besten passt..
Alle Lösungen sind immer nicht hinreichend zufriedenstellend.
Erste kunde: Die Kreisbahn
//teilt einen gedachten kreis(Radius) in seqmente und erzeugt eine //Position für jedes seqment, kreispunktbahn void Circle::Filter(CRect rc,int DotCnt,double Radius) { #define FC (2.0f*PI/(double)SEGMENTS) CPoint ptc(rc.CenterPoint()); double r(rc.Width()/2); for(register short nr=0; nr<SEGMENTS; nr++)//calc dotring { int x = ptc.x + cos(FC*nr) * r; int y = ptc.y + sin(FC*nr) * r; } }
Zweite Kunde: Bresenhalm Kreis !
oid CGdiObj::Circle(CDC *pDC, WORD x0, WORD y0, WORD radius) {//breshalm :-) short f = 1 - radius; WORD ddFx = 0,ddFy = -2 * radius; WORD x = 0, y = radius; while(x < y) { if(f >= 0) y--, ddFy+=2, f+=ddFy; x++, ddFx+=2, f+=ddFx+1; pDC->SetPixel(x0 + x, y0 + y,RGB(0,255,0)); pDC->SetPixel(x0 - x, y0 + y,RGB(0,255,0)); pDC->SetPixel(x0 + x, y0 - y,RGB(0,255,0)); pDC->SetPixel(x0 - x, y0 - y,RGB(0,255,0)); pDC->SetPixel(x0 + y, y0 + x,RGB(0,255,0)); pDC->SetPixel(x0 - y, y0 + x,RGB(0,255,0)); pDC->SetPixel(x0 + y, y0 - x,RGB(0,255,0)); pDC->SetPixel(x0 - y, y0 - x,RGB(0,255,0)); } }
Dritte Kunde Kreis aus drei punkten
bool Circle::Calc(double ScaleX, double ScaleY,double &PosX, double &PosY, double &Rad) { // Die Unbekannten: X, Y und Radius mit Naeherungswerten double X=0,Y=0,R=1; unsigned int pointCnt = m_Points.size(); for(unsigned int iteration=0; iteration < pointCnt; iteration++) { GdiMatrix A(pointCnt, 3); GdiMatrix b(pointCnt, 1); for(unsigned int i=0; i<pointCnt; i++) { GdiPoint &p = m_Points[i]; p.x *= ScaleX; p.y *= ScaleY; // X X+= DX; double px=GdiPoint(X,Y).Cross(p)-R; X-=2.0*DX; double mx=GdiPoint(X,Y).Cross(p)-R; X+= DX; // Y Y+= DX; double py=GdiPoint(X,Y).Cross(p)-R; Y-=2.0*DX; double my=GdiPoint(X,Y).Cross(p)-R; Y+= DX; A.SetElement(i,0,(px-mx)/(2.0*DX)); A.SetElement(i,1,(py-my)/(2.0*DX)); A.SetElement(i,2,-1); double val = GdiPoint(X,Y).Cross(p)-R; b.SetElement(i,0,-val); } GdiMatrix tmpA = A.GetTransposed()*A; GdiMatrix tmpB = A.GetTransposed()*b; GdiMatrix dx = tmpA.GetInverted() * tmpB; X = X+dx.GetElement(0,0); Y = Y+dx.GetElement(1,0); R = R+dx.GetElement(2,0); } PosX = X, PosY = Y , Rad = R; return true; }
Vierte Kunde: Matrix Interpolation mit irregulärer Matrix (Kreis aus drei punkten Zweite Version)
bool Circle::Info(GdiPoint a,GdiPoint b,GdiPoint c,int &PosX, int &PosY, int &Rad) { #define COLS 4 #define ROWS 3 double *pM,*pMat = new double[COLS*ROWS]; if(!(pM=pMat)) return false; pM[0] = 1,pM[1] = -2*a.x, pM[2] = -2*a.y, pM[3] = -a.x*a.x-a.y*a.y; pM+=COLS; pM[0] = 1,pM[1] = -2*b.x, pM[2] = -2*b.y, pM[3] = -b.x*b.x-b.y*b.y; pM+=COLS; pM[0] = 1,pM[1] = -2*c.x, pM[2] = -2*c.y, pM[3] = -c.x*c.x-c.y*c.y; pM+=COLS; for(register int j = 0; j < ROWS; j++) { // Diagonalenfeld normalisieren register double q = pMat[j * COLS + j]; if(q == 0)//Gewährleisten, daß keine 0 in der Diagonale steht for(register int i = j + 1; i < ROWS; i++) if(pMat[i * COLS + j] != 0)// Suche Reihe mit Feld <> 0 und addiere dazu { for(register int k = 0 ; k < COLS; k++) pMat[j * COLS + k] += pMat[i * COLS + k]; q = pMat[j * COLS + j]; break; } if(q != 0)// Diagonalen auf 1 bringen for(register int k = 0; k < COLS; k++) pMat[j * COLS + k] = pMat[j * COLS + k] / q; for(register int i = 0 ; i < ROWS; i++) if(i != j )// Spalten außerhalb der Diagonalen auf 0 bringen { q = pMat[i * COLS + j]; for(register int k = 0; k < COLS; k++) pMat[i * COLS + k] -= q * pMat[j * COLS + k]; } } bool stat(pMat[0] != 1 || pMat[5] != 1 || pMat[10] != 1); if (stat)//irrationale matrix { PosX = pMat[7]; PosY = pMat[11]; Rad = (sqrt(PosX*PosX + PosY*PosY - pMat[3])); } delete pMat; return stat; }
Fünfte Kunde: Die spline Line
bringt einen auch leicht in die klapse.#ifndef _SPLINE_H #define _SPLINE_H #define DIV_FACTOR 4 //adjust this factor to adjust the curve smoothness class Curve { public: float Ax,Ay; float Bx,By; float Cx,Cy; int Ndiv; Curve(float ax,float ay,float bx,float by,float cx,float cy,int ndiv) { Ax = ax; Ay = ay; Bx = bx; By = by; Cx = cx; Cy = cy; Ndiv = ndiv; } Curve(float ax,float ay,float bx,float by,float cx,float cy) { Ax = ax; Ay = ay; Bx = bx; By = by; Cx = cx; Cy = cy; Ndiv = (int)(max(abs((int)Ax),abs((int)Ay))/DIV_FACTOR); } Curve() { }; void PutCurve(float ax,float ay,float bx,float by,float cx,float cy) { Ax = ax; Ay = ay; Bx = bx; By = by; Cx = cx; Cy = cy; Ndiv = (int)(max(abs((int)Ax),abs((int)Ay))/DIV_FACTOR); } void draw(HDC hdc,float x,float y) { int OrigX,OrigY,NewX,NewY; float t,f,g,h; if (Ndiv==0) Ndiv=1; OrigX = (int)x; OrigY= (int)y; for(int i=1; i<=Ndiv ; i++) { t = 1.0f / (float)Ndiv * (float)i; f = t*t*(3.0f-2.0f*t); g = t*(t-1.0f)*(t-1.0f); h = t*t*(t-1.0f); NewX = (int)(x + Ax*f + Bx*g + Cx*h); NewY = (int)(y + Ay*f + By*g + Cy*h); MoveToEx(hdc, OrigX, OrigY, NULL); LineTo(hdc, NewX, NewY); OrigX = NewX; OrigY=NewY; } } int GetCount() { if (Ndiv==0) Ndiv=1; int PointCount = 1; for(int i=1; i<=Ndiv ; i++) { PointCount++; } return PointCount; } void GetCurve(float x,float y, POINT points[], int& PointCount) { int X,Y; float t,f,g,h; if (Ndiv==0) Ndiv=1; X = (int)x; Y= (int)y; points[PointCount].x = X; points[PointCount].y = Y; PointCount++; for(int i=1; i<=Ndiv ; i++) { t = 1.0f / (float)Ndiv * (float)i; f = t*t*(3.0f-2.0f*t); g = t*(t-1.0f)*(t-1.0f); h = t*t*(t-1.0f); X = (int)(x + Ax*f + Bx*g + Cx*h); Y = (int)(y + Ay*f + By*g + Cy*h); points[PointCount].x = X; points[PointCount].y = Y; PointCount++; } } }; class Spline { public: float* Px; float* Py; float* Ax; float* Ay; float* Bx; float* By; float* Cx; float* Cy; float* k; float* Mat[3]; int NP; // constructor Spline(POINT pt[], int np) { NP = np; Px = new float[NP]; Py = new float[NP]; Ax = new float[NP]; Ay = new float[NP]; Bx = new float[NP]; By = new float[NP]; Cx = new float[NP]; Cy = new float[NP]; k = new float[NP]; Mat[0] = new float[NP]; Mat[1] = new float[NP]; Mat[2] = new float[NP]; for(int i=0;i<NP ;i++) { Px[i] = (float)pt[i].x; Py[i] = (float)pt[i].y; } } Spline(float px[] , float py[] , int np) { NP = np; Px = new float[NP]; Py = new float[NP]; Ax = new float[NP]; Ay = new float[NP]; Bx = new float[NP]; By = new float[NP]; Cx = new float[NP]; Cy = new float[NP]; k = new float[NP]; Mat[0] = new float[NP]; Mat[1] = new float[NP]; Mat[2] = new float[NP]; for(int i=0;i<NP ;i++) { Px[i] = px[i]; Py[i] = py[i]; } } ~Spline() { delete[] Px; delete[] Py; delete[] Ax; delete[] Ay; delete[] Bx; delete[] By; delete[] Cx; delete[] Cy; delete[] k; delete[] Mat[0]; delete[] Mat[1]; delete[] Mat[2]; } void Generate() { float AMag , AMagOld; // vector A for(int i= 0 ; i<=NP-2 ; i++ ) { Ax[i] = Px[i+1] - Px[i]; Ay[i] = Py[i+1] - Py[i]; } // k AMagOld = (float)sqrt(Ax[0]*Ax[0] + Ay[0]*Ay[0]); for(register int i=0 ; i<=NP-3 ; i++) { AMag = (float)sqrt(Ax[i+1]*Ax[i+1] + Ay[i+1]*Ay[i+1]); k[i] = AMagOld / AMag; AMagOld = AMag; } k[NP-2] = 1.0f; // Matrix for(register int i=1; i<=NP-2;i++) { Mat[0][i] = 1.0f; Mat[1][i] = 2.0f*k[i-1]*(1.0f + k[i-1]); Mat[2][i] = k[i-1]*k[i-1]*k[i]; } Mat[1][0] = 2.0f; Mat[2][0] = k[0]; Mat[0][NP-1] = 1.0f; Mat[1][NP-1] = 2.0f*k[NP-2]; // for(register int i=1; i<=NP-2;i++) { Bx[i] = 3.0f*(Ax[i-1] + k[i-1]*k[i-1]*Ax[i]); By[i] = 3.0f*(Ay[i-1] + k[i-1]*k[i-1]*Ay[i]); } Bx[0] = 3.0f*Ax[0]; By[0] = 3.0f*Ay[0]; Bx[NP-1] = 3.0f*Ax[NP-2]; By[NP-1] = 3.0f*Ay[NP-2]; // MatrixSolve(Bx); MatrixSolve(By); for(register int i=0 ; i<=NP-2 ; i++ ) { Cx[i] = k[i]*Bx[i+1]; Cy[i] = k[i]*By[i+1]; } } void MatrixSolve(float B[]) { float* Work = new float[NP]; float* WorkB = new float[NP]; for(int i=0;i<=NP-1;i++) { Work[i] = B[i] / Mat[1][i]; WorkB[i] = Work[i]; } for(int j=0 ; j<10 ; j++) { /// need convergence judge Work[0] = (B[0] - Mat[2][0]*WorkB[1])/Mat[1][0]; for(int i=1; i<NP-1 ; i++ ) { Work[i] = (B[i]-Mat[0][i]*WorkB[i-1]-Mat[2][i]*WorkB[i+1]) /Mat[1][i]; } Work[NP-1] = (B[NP-1] - Mat[0][NP-1]*WorkB[NP-2])/Mat[1][NP-1]; for(register int i=0 ; i<=NP-1 ; i++ ) { WorkB[i] = Work[i]; } } for(register int i=0 ; i<=NP-1 ; i++ ) { B[i] = Work[i]; } delete[] Work; delete[] WorkB; } void draw(HDC hdc) { Curve c; for(int i=0; i<NP-1 ; i++) { c.PutCurve(Ax[i],Ay[i],Bx[i],By[i],Cx[i],Cy[i]); c.draw(hdc,Px[i],Py[i]); } } int GetCurveCount() { Curve c; int count = 0; for(int i=0; i<NP-1 ; i++) { c.PutCurve(Ax[i],Ay[i],Bx[i],By[i],Cx[i],Cy[i]); count += c.GetCount(); } return count; } void GetCurve(POINT points[], int& PointCount) { Curve c; for(int i=0; i<NP-1 ; i++) { c.PutCurve(Ax[i],Ay[i],Bx[i],By[i],Cx[i],Cy[i]); c.GetCurve(Px[i],Py[i], points, PointCount); } } //////////// closed cubic spline //////////////////// void GenClosed() { float AMag , AMagOld , AMag0; // vector A for(int i= 0 ; i<=NP-2 ; i++ ) { Ax[i] = Px[i+1] - Px[i]; Ay[i] = Py[i+1] - Py[i]; } Ax[NP-1] = Px[0] - Px[NP-1]; Ay[NP-1] = Py[0] - Py[NP-1]; // k AMag0 = AMagOld = (float)sqrt(Ax[0]*Ax[0] + Ay[0]*Ay[0]); for(register int i=0 ; i<=NP-2 ; i++) { AMag = (float)sqrt(Ax[i+1]*Ax[i+1] + Ay[i+1]*Ay[i+1]); k[i] = AMagOld / AMag; AMagOld = AMag; } k[NP-1]=AMagOld/AMag0; // Matrix for(register int i=1; i<=NP-1;i++) { Mat[0][i] = 1.0f; Mat[1][1] = 2.0f*k[i-1]*(1.0f + k[i-1]); Mat[2][i] = k[i-1]*k[i-1]*k[i]; } Mat[0][0] = 1.0f; Mat[1][0] = 2.0f*k[NP-1]*(1.0f + k[NP-1]); Mat[2][0] = k[NP-1]*k[NP-1]*k[0]; // for(register int i=1; i<=NP-1;i++) { Bx[i] = 3.0f*(Ax[i-1] + k[i-1]*k[i-1]*Ax[i]); By[i] = 3.0f*(Ay[i-1] + k[i-1]*k[i-1]*Ay[i]); } Bx[0] = 3.0f*(Ax[NP-1] + k[NP-1]*k[NP-1]*Ax[0]); By[0] = 3.0f*(Ay[NP-1] + k[NP-1]*k[NP-1]*Ay[0]); // MatrixSolveEX(Bx); MatrixSolveEX(By); for(register int i=0 ; i<=NP-2 ; i++ ) { Cx[i] = k[i]*Bx[i+1]; Cy[i] = k[i]*By[i+1]; } Cx[NP-1] = k[NP-1]*Bx[0]; Cy[NP-1] = k[NP-1]*By[0]; } ///// tridiagonal matrix + elements of [0][0], [N-1][N-1] //// void MatrixSolveEX(float B[]) { float* Work = new float[NP]; float* WorkB = new float[NP]; for(int i=0;i<=NP-1;i++) { Work[i] = B[i] / Mat[1][i]; WorkB[i] = Work[i]; } for(int j=0 ; j<10 ; j++) { // need judge of convergence Work[0] = (B[0]-Mat[0][0]*WorkB[NP-1]-Mat[2][0]*WorkB[1]) /Mat[1][0]; for(int i=1; i<NP-1 ; i++ ) { Work[i] = (B[i]-Mat[0][i]*WorkB[i-1]-Mat[2][i]*WorkB[i+1]) /Mat[1][i]; } Work[NP-1] = (B[NP-1]-Mat[0][NP-1]*WorkB[NP-2]-Mat[2][NP-1]*WorkB[0]) /Mat[1][NP-1]; for(register int i=0 ; i<=NP-1 ; i++ ) { WorkB[i] = Work[i]; } } for(register int i=0 ; i<=NP-1 ; i++ ) { B[i] = Work[i]; } delete[] Work; delete[] WorkB; } void drawClosed(HDC hdc) { Curve c; for(int i=0; i<NP ; i++) { c.PutCurve(Ax[i],Ay[i],Bx[i],By[i],Cx[i],Cy[i]); c.draw(hdc ,Px[i],Py[i]); } } }; #endif
Hier noch der benötigte Matrix Stuff:
Header für Matrix
class GdiMatrix { public: public: friend class MatrixHelper ; // used for operator[][] // construction and destruction GdiMatrix() ; // default constructor GdiMatrix(const GdiMatrix &other) ; // copy constructor GdiMatrix(int nCols, int nRows) ; // constructs an empty matrix of this size GdiMatrix(int size, bool set_diagonal = true) ; // creates a square matrix // GdiMatrix(VARIANT& var) ; // from a SAFEARRAY variant virtual ~GdiMatrix(); // destructor // matrix mathematical operations GdiMatrix& operator=(const GdiMatrix &other) ; GdiMatrix operator+(const GdiMatrix &other) const ; GdiMatrix operator-(const GdiMatrix &other) const ; GdiMatrix operator*(const GdiMatrix &other) const ; void operator+=(const GdiMatrix &other) ; void operator-=(const GdiMatrix &other) ; void operator*=(const GdiMatrix &other) ; void operator*=(double value) ; friend GdiMatrix operator*(const GdiMatrix &other, double value) ; bool operator==(const GdiMatrix &other) const ; const MatrixHelper operator[](int nCol) const ; // reading version MatrixHelper operator[](int nCol) ; // writing version // element access bool SetElement(int nCol, int nRow, double value) ; #ifdef _DEBUG double GetElement(int nCol, int nRow) const ; #else inline double GetElement(int nCol, int nRow) const { return m_pData[nCol + nRow * m_NumColumns] ; } ; #endif inline int GetNumColumns() const { return m_NumColumns ; } ; inline int GetNumRows() const { return m_NumRows ; } ; double SumColumn(int col) const ; double SumRow(int row) const ; double SumColumnSquared(int col) const ; double SumRowSquared(int row) const ; double GetRowMin(int row) const ; double GetRowMax(int row) const ; double GetColumnMin(int col) const ; double GetColumnMax(int col) const ; // matrix transposition GdiMatrix GetTransposed() const ; void Transpose() ; // matrix inversion GdiMatrix GetInverted() const ; void Invert() ; // covariant (A' * A) GdiMatrix Covariant() const ; // normalisation GdiMatrix GetNormalised(double min, double max) const ; void Normalise(double min, double max) ; // ranges functions void GetNumericRange(double &min, double &max) const ; // matrix concatenation GdiMatrix GetConcatinatedColumns(const GdiMatrix& other) const ; void ConcatinateColumns(const GdiMatrix &other) ; GdiMatrix GetConcatinatedRows(const GdiMatrix& other) const ; void ConcatinateRows(const GdiMatrix &other) ; // adds an new row / column to the matrix void AddColumn(const double *pData) ; void AddRow(const double *pData) ; // sub matrix extraction, setting GdiMatrix ExtractSubMatrix(int col_start, int row_start, int col_size, int row_size) const ; void SetSubMatrix(int col_start, int row_start, const GdiMatrix &other) ; GdiMatrix ExtractDiagonal() const ; // squaring the matrix functions GdiMatrix GetSquareMatrix() const ; void MakeSquare() ; // export functions/import void CopyToClipboard() const ; private: // internal variables int m_NumColumns ; // number of columns in matrix int m_NumRows ; // number of rows in matrix double* m_pData ; // pointer to data, may be shared among objects #ifdef _DEBUG // variables used in debug for obejct counting static int m_NextObjectNumber ; int m_ObjectNumber ; #endif private: // private internal functions double* AllocateMemory(int nCols, int nROws) ; // reference counting functions void IncrementReferenceCount() ; // increments the m_pData reference count void DecrementReferenceCount() ; // decrements the m_pData reference count void DecrementAndRelease() ; // decrements the count and releases the memory if required int GetReferenceCount() const ; // returns the m_pData's reference count // helper functions CString GetRowAsText(int row) const ; static CString ReadLine(CFile &file) ; // reads a \r\n delimited line of text from a file static int GetStringToken(CString source, CString &destination, int start, char ch) ; }; // this class is used to help the operator[][] on a matrix object work correctly // it only provides the operator[] class MatrixHelper { public: friend class GdiMatrix ; protected: // protected constructor so only friend class can construct // a MatrixHelper object MatrixHelper(GdiMatrix* pMatrix, int col) : m_pMatrix(pMatrix), m_pMatrixConst(NULL) { m_Col = col ; } ; MatrixHelper(const GdiMatrix* const pMatrix, int col) : m_pMatrixConst(pMatrix), m_pMatrix(NULL) { m_Col = col ; } ; MatrixHelper(MatrixHelper& other) : m_pMatrix(other.m_pMatrix), m_pMatrixConst(other.m_pMatrixConst) { m_Col = other.m_Col ; } ; public: ~MatrixHelper() { } ; double operator[](int row) const { ASSERT(row >= 0) ; // array bounds error ASSERT(row < m_pMatrixConst->m_NumRows) ; // array bounds error return m_pMatrixConst->m_pData[row * m_pMatrixConst->m_NumColumns + m_Col] ; } ; double& operator[](int row) { // first check the reference count on our data object to see whether we need to create a copy if (m_pMatrix->GetReferenceCount() > 1) { // we need to make a copy double *pData = m_pMatrix->m_pData ; // take a copy of the pointer m_pMatrix->DecrementReferenceCount() ; // decrement the current reference count m_pMatrix->m_pData = m_pMatrix->AllocateMemory(m_pMatrix->m_NumColumns, m_pMatrix->m_NumRows) ; memcpy(m_pMatrix->m_pData, pData, sizeof(double) * m_pMatrix->m_NumColumns * m_pMatrix->m_NumRows) ; m_pMatrix->IncrementReferenceCount() ; // increment the new data's reference count } ASSERT(row >= 0) ; // array bounds error ASSERT(row < m_pMatrix->m_NumRows) ; // array bounds error ASSERT(m_pMatrix->m_pData) ; return m_pMatrix->m_pData[m_Col + row * m_pMatrix->m_NumColumns] ; } ; private: GdiMatrix* m_pMatrix ; const GdiMatrix* m_pMatrixConst ; int m_Col ; // column index for operator[][] } ;
Matrix Code
#include "stdafx.h" #include "Matrix.h" GdiMatrix::GdiMatrix() { // default constructor, create a 1 * 1 array m_NumColumns = 1 ; m_NumRows = 1 ; m_pData = NULL ; m_pData = AllocateMemory(m_NumColumns, m_NumRows) ; IncrementReferenceCount() ; // count the reference to this memory } GdiMatrix::GdiMatrix(const GdiMatrix &other) { // copy constructor m_pData = NULL ; // use the other objects data pointer m_NumColumns = other.m_NumColumns ; m_NumRows = other.m_NumRows ; m_pData = other.m_pData ; // copy the pointer IncrementReferenceCount() ; // this thread can get the mutex multiple times without blocking } GdiMatrix::GdiMatrix(int nCols, int nRows) { // size constructor ASSERT(nCols > 0) ; // matrix size error ASSERT(nRows > 0) ; // matrix size error m_pData = NULL ; m_NumColumns = nCols ; m_NumRows = nRows ; m_pData = AllocateMemory(m_NumColumns, m_NumRows) ; IncrementReferenceCount() ; // count the reference to this memory } GdiMatrix::GdiMatrix(int size, bool set_diagonal) { // size constructor ASSERT(size > 0) ; // matrix size error m_pData = NULL ; m_NumColumns = size ; m_NumRows = size ; m_pData = AllocateMemory(m_NumColumns, m_NumRows) ; IncrementReferenceCount() ; // count the reference to this memory // set the dialognal if required if (set_diagonal) { for (int i = 0 ; i < size ; ++i) SetElement(i, i, 1.0) ; } } GdiMatrix::~GdiMatrix() { #ifdef _DEBUG TRACE("Destroying GdiMatrix object %1d\n", m_ObjectNumber) ; #endif DecrementAndRelease() ; // free's m_pData if no other references } double* GdiMatrix::AllocateMemory(int nCols, int nRows) { ASSERT(nCols > 0) ; // size error ASSERT(nRows > 0) ; // size error // allocates heap memory for an array double *pData = NULL ; pData = new double[nCols * nRows + 1] ; // all in one allocation (+1 for reference count) ASSERT(pData != NULL) ; // allocation error ASSERT(FALSE == IsBadReadPtr(pData, sizeof(double) * (nCols * nRows + 1))) ; // empty the memory memset(pData, 0, sizeof(double) * (nCols * nRows + 1)) ; // starts with a 0 reference count return pData ; } GdiMatrix& GdiMatrix::operator=(const GdiMatrix &other) { if (&other == this) return *this ; // this does the same job as a copy constructor except we have to de-allocate any // memory we may have already allocated DecrementAndRelease() ; // free's m_pData if no other references // now copy the other matrix into ourselves // use the other objects data pointer m_NumColumns = other.m_NumColumns ; m_NumRows = other.m_NumRows ; m_pData = other.m_pData ; // copy the pointer IncrementReferenceCount() ; // this thread can get the mutex multiple times without blocking // finally return a reference to ourselves return *this ; } bool GdiMatrix::operator==(const GdiMatrix &other) const { // only return true if the matrices are exactly the same if (&other == this) return true ; // comparing to ourselves if (m_pData == other.m_pData) return true ; // both pointing to same data, must be same if (m_NumColumns != other.m_NumColumns || m_NumRows != other.m_NumRows) return false ; // different dimensions if (memcmp(m_pData, other.m_pData, sizeof(double) * m_NumColumns * m_NumRows) == 0) return true ; // buffers are the same return false ; // must be different } GdiMatrix GdiMatrix::operator+(const GdiMatrix &other) const { // first check for a valid addition operation if (m_NumColumns != other.m_NumColumns) throw "Invalid operation" ; if (m_NumRows != other.m_NumRows) throw "Invalid operation" ; // now that we know that the operation is possible ASSERT(FALSE == IsBadReadPtr(other.m_pData, sizeof(double) * other.m_NumColumns * other.m_NumRows)) ; // construct the object we are going to return GdiMatrix result(*this) ; // copy ourselves // now add in the other matrix for (int i = 0 ; i < m_NumColumns ; ++i) { for (int j = 0 ; j < m_NumRows ; ++j) result.SetElement(i, j, result.GetElement(i, j) + other.GetElement(i, j)) ; } return result ; } GdiMatrix GdiMatrix::operator-(const GdiMatrix &other) const { // first check for a valid subtraction operation if (m_NumColumns != other.m_NumColumns) throw "Invalid operation" ; if (m_NumRows != other.m_NumRows) throw "Invalid operation" ; // now that we know that the operation is possible ASSERT(FALSE == IsBadReadPtr(other.m_pData, sizeof(double) * other.m_NumColumns * other.m_NumRows)) ; // construct the object we are going to return GdiMatrix result(*this) ; // copy ourselves // now subtract the other matrix for (int i = 0 ; i < m_NumColumns ; ++i) { for (int j = 0 ; j < m_NumRows ; ++j) result.SetElement(i, j, result.GetElement(i, j) - other.GetElement(i, j)) ; } return result ; } GdiMatrix GdiMatrix::operator*(const GdiMatrix &other) const { // first check for a valid multiplication operation if (m_NumRows != other.m_NumColumns) throw "Matrices do not have common size" ; // now that we know that the operation is possible ASSERT(FALSE == IsBadReadPtr(other.m_pData, sizeof(double) * other.m_NumColumns * other.m_NumRows)) ; // construct the object we are going to return GdiMatrix result(m_NumColumns, other.m_NumRows) ; // e.g. // [A][B][C] [G][H] [A*G + B*I + C*K][A*H + B*J + C*L] // [D][E][F] * [I][J] = [D*G + E*I + F*K][D*H + E*J + F*L] // [K][L] // double value ; for (int i = 0 ; i < result.m_NumColumns ; ++i) { for (int j = 0 ; j < result.m_NumRows ; ++j) { value = 0.0 ; for (int k = 0 ; k < m_NumRows ; ++k) { value += GetElement(i, k) * other.GetElement(k, j) ; } result.SetElement(i, j, value) ; } } return result ; } void GdiMatrix::operator+=(const GdiMatrix &other) { // first check for a valid addition operation if (m_NumColumns != other.m_NumColumns) throw "Invalid operation" ; if (m_NumRows != other.m_NumRows) throw "Invalid operation" ; // now that we know that the operation is possible ASSERT(FALSE == IsBadReadPtr(other.m_pData, sizeof(double) * other.m_NumColumns * other.m_NumRows)) ; // now add in the other matrix for (int i = 0 ; i < m_NumColumns ; ++i) { for (int j = 0 ; j < m_NumRows ; ++j) SetElement(i, j, GetElement(i, j) + other.GetElement(i, j)) ; } } void GdiMatrix::operator-=(const GdiMatrix &other) { // first check for a valid subtraction operation if (m_NumColumns != other.m_NumColumns) throw "Invalid operation" ; if (m_NumRows != other.m_NumRows) throw "Invalid operation" ; // now that we know that the operation is possible ASSERT(FALSE == IsBadReadPtr(other.m_pData, sizeof(double) * other.m_NumColumns * other.m_NumRows)) ; // now subtract the other matrix for (int i = 0 ; i < m_NumColumns ; ++i) { for (int j = 0 ; j < m_NumRows ; ++j) SetElement(i, j, GetElement(i, j) - other.GetElement(i, j)) ; } } void GdiMatrix::operator*=(const GdiMatrix &other) { // first check for a valid multiplication operation if (m_NumRows != other.m_NumColumns) throw "Matrices do not have common size" ; *this = *this * other ; } void GdiMatrix::operator*=(double value) { // just multiply the elements by the value for (int i = 0 ; i < m_NumColumns ; ++i) { for (int j = 0 ; j < m_NumRows ; ++j) { SetElement(i, j, GetElement(i, j) * value) ; } } } // MatrixHelper is only used for this to simulate a GdiMatrix::operator[][] const MatrixHelper GdiMatrix::operator[](int nCol) const { ASSERT(nCol >= 0) ; // array bounds error ASSERT(nCol < m_NumColumns) ; // array bounds error // construc the MatrixHelper object to allow operator[][] to work MatrixHelper mh(this, nCol) ; return mh ; } MatrixHelper GdiMatrix::operator[](int nCol) { ASSERT(nCol >= 0) ; // array bounds error ASSERT(nCol < m_NumColumns) ; // array bounds error // construc the MatrixHelper object to allow operator[][] to work MatrixHelper mh(this, nCol) ; return mh ; } bool GdiMatrix::SetElement(int nCol, int nRow, double value) { // first check the reference count on our data object to see whether we need to create a copy if (GetReferenceCount() > 1) { // we need to make a copy double *pData = m_pData ; // take a copy of the pointer DecrementReferenceCount() ; // decrement the current reference count m_pData = AllocateMemory(m_NumColumns, m_NumRows) ; memcpy(m_pData, pData, sizeof(double) * m_NumColumns * m_NumRows) ; IncrementReferenceCount() ; // increment the new data's reference count } ASSERT(nCol >= 0) ; // array bounds error ASSERT(nCol < m_NumColumns) ; // array bounds error ASSERT(nRow >= 0) ; // array bounds error ASSERT(nRow < m_NumRows) ; // array bounds error ASSERT(m_pData) ; // bad pointer error m_pData[nCol + nRow * m_NumColumns] = value ; return true ; } #ifdef _DEBUG // release version is in-line double GdiMatrix::GetElement(int nCol, int nRow) const { ASSERT(nCol >= 0) ; // array bounds error ASSERT(nCol < m_NumColumns) ; // array bounds error ASSERT(nRow >= 0) ; // array bounds error ASSERT(nRow < m_NumRows) ; // array bounds error ASSERT(m_pData) ; // bad pointer error return m_pData[nCol + nRow * m_NumColumns] ; } #endif // // To avoid big hits when constructing and assigning GdiMatrix objects, multiple GdiMatrix's can reference // the same m_pData member. Only when a matrix becomes different from the other does a new version of the array // get created and worked with. // void GdiMatrix::IncrementReferenceCount() { // get a pointer to the end of the m_pData object where the reference count resides int* pReference = (int*)&m_pData[m_NumColumns * m_NumRows] ; ++(*pReference) ; // increment the reference count // done! } void GdiMatrix::DecrementReferenceCount() { // get a pointer to the end of the m_pData object where the reference count resides int* pReference = (int*)&m_pData[m_NumColumns * m_NumRows] ; --(*pReference) ; // decrement the reference count // done! } void GdiMatrix::DecrementAndRelease() { // get a pointer to the end of the m_pData object where the reference count resides int* pReference = (int*)&m_pData[m_NumColumns * m_NumRows] ; --(*pReference) ; // decrement the reference count if (*pReference == 0) { // the memory is no longer needed, release it delete []m_pData ; m_pData = NULL ; } // done! } int GdiMatrix::GetReferenceCount() const { // get a pointer to the end of the m_pData object where the reference count resides int* pReference = (int*)&m_pData[m_NumColumns * m_NumRows] ; return *pReference ; } GdiMatrix GdiMatrix::GetTransposed() const { GdiMatrix transposed(*this) ; // make a copy of ourselves transposed.Transpose() ; return transposed ; } void GdiMatrix::Transpose() { // first check the reference count on our data object to see whether we need to create a copy GdiMatrix mcopy(*this) ; // swap the x/y values int copy = m_NumColumns ; m_NumColumns = m_NumRows ; m_NumRows = copy ; // copy across the transposed data for (int i = 0 ; i < m_NumColumns ; ++i) { for (int j = 0 ; j < m_NumRows ; ++j) SetElement(i, j, mcopy.GetElement(j, i)) ; } } GdiMatrix GdiMatrix::GetInverted() const { // matrix inversion will only work on square matrices if (m_NumColumns != m_NumRows) throw "GdiMatrix must be square." ; // return this matrix inverted GdiMatrix copy(*this) ; copy.Invert() ; return copy ; } void GdiMatrix::Invert() { // matrix inversion will only work on square matrices // invert ourselves if (m_NumColumns != m_NumRows) throw "GdiMatrix must be square." ; double e ; for (int k = 0 ; k < m_NumColumns ; ++k) { e = GetElement(k, k) ; SetElement(k, k, 1.0) ; if (e == 0.0) break;//throw "GdiMatrix inversion error" ; for (int j = 0 ; j < m_NumColumns ; ++j) SetElement(k, j, GetElement(k, j) / e) ; for (int i = 0 ; i < m_NumColumns ; ++i) { if (i != k) { e = GetElement(i, k) ; SetElement(i, k, 0.0) ; for (int j = 0 ; j < m_NumColumns ; ++j) SetElement(i, j, GetElement(i, j) - e * GetElement(k, j)) ; } } } } // A' * A GdiMatrix GdiMatrix::Covariant() const { GdiMatrix result ; GdiMatrix trans(GetTransposed()) ; result = *this * trans ; return result ; } GdiMatrix GdiMatrix::ExtractSubMatrix(int col_start, int row_start, int col_size, int row_size) const { ASSERT(col_start >= 0) ; // bad start index ASSERT(row_start >= 0) ; // bad start index ASSERT(col_size > 0) ; // bad size ASSERT(row_size > 0) ; // bad size // make sure the requested sub matrix is in the current matrix if (col_start + col_size > m_NumColumns) throw "Sub matrix is not contained in source" ; if (row_start + row_size > m_NumRows) throw "Sub matrix is not contained in source" ; GdiMatrix sub(col_size, row_size) ; for (int i = 0 ; i < col_size ; ++i) { for (int j = 0 ; j < row_size ; ++j) { sub.SetElement(i, j, GetElement(col_start + i, row_start + j)) ; } } return sub ; } void GdiMatrix::SetSubMatrix(int col_start, int row_start, const GdiMatrix &other) { ASSERT(col_start >= 0) ; // bad start index ASSERT(row_start >= 0) ; // bad start index ASSERT(col_start + other.m_NumColumns <= m_NumColumns) ; // bad size ASSERT(row_start + other.m_NumRows <= m_NumRows) ; // bad size for (int i = 0 ; i < other.m_NumColumns ; ++i) { for (int j = 0 ; j < other.m_NumRows ; ++j) { SetElement(col_start + i, row_start + j, other.GetElement(i, j)) ; } } } GdiMatrix GdiMatrix::ExtractDiagonal() const { if (m_NumColumns != m_NumRows) throw "Can only extract diagonal from square matrix" ; GdiMatrix diagonal(m_NumColumns, 1) ; for (int i = 0 ; i < m_NumColumns ; ++i) diagonal.SetElement(i, 0, GetElement(i, i)) ; return diagonal ; } GdiMatrix GdiMatrix::GetConcatinatedColumns(const GdiMatrix& other) const { if (m_NumRows != other.m_NumRows) throw "Cannot concatenate matrices, not same size" ; // copy ourselves and then return the concatenated result GdiMatrix result(*this) ; result.ConcatinateColumns(other) ; return result ; } // concatinate the other matrix to ourselves void GdiMatrix::ConcatinateColumns(const GdiMatrix &other) { if (m_NumRows != other.m_NumRows) throw "Cannot concatenate matrices, not same size" ; // create a matrix big enough to hold both GdiMatrix result(m_NumColumns + other.m_NumColumns, m_NumRows) ; // now populate it for (int i = 0 ; i < m_NumColumns ; ++i) { for (int j = 0 ; j < m_NumRows ; ++j) { result.SetElement(i, j, GetElement(i, j)) ; } } // now add the other matrix for (int i = 0 ; i < other.m_NumColumns ; ++i) { for (int j = 0 ; j < m_NumRows ; ++j) { result.SetElement(i + m_NumColumns, j, other.GetElement(i, j)) ; } } *this = result ; // assign it to us } GdiMatrix GdiMatrix::GetConcatinatedRows(const GdiMatrix& other) const { if (m_NumColumns != other.m_NumColumns) throw "Cannot concatenate matrices, not same size" ; // copy ourselves and then return the concatenated result GdiMatrix result(*this) ; result.ConcatinateRows(other) ; return result ; } void GdiMatrix::ConcatinateRows(const GdiMatrix &other) { if (m_NumColumns != other.m_NumColumns) throw "Cannot concatenate matrices, not same size" ; // create a matrix big enough to hold both GdiMatrix result(m_NumColumns, m_NumRows + other.m_NumRows) ; // now populate it for (int i = 0 ; i < m_NumColumns ; ++i) { for (int j = 0 ; j < m_NumRows ; ++j) { result.SetElement(i, j, GetElement(i, j)) ; } } // now add the other matrix for (int i = 0 ; i < other.m_NumColumns ; ++i) { for (int j = 0 ; j < m_NumRows ; ++j) { result.SetElement(i, j + m_NumRows, other.GetElement(i, j)) ; } } *this = result ; // assign it to us } void GdiMatrix::AddColumn(const double *pData) { ASSERT(FALSE == IsBadReadPtr(pData, sizeof(double) * m_NumRows)) ; GdiMatrix result(m_NumColumns + 1, m_NumRows) ; // costruct the result result.SetSubMatrix(0, 0, *this) ; // copy ouselves across // now add the new row for (int i = 0 ; i < m_NumRows ; ++i) { result.SetElement(m_NumColumns, i, pData[i]) ; } *this = result ; // assign result to us } void GdiMatrix::AddRow(const double *pData) { ASSERT(FALSE == IsBadReadPtr(pData, sizeof(double) * m_NumColumns)) ; GdiMatrix result(m_NumColumns, m_NumRows + 1) ; // costruct the result result.SetSubMatrix(0, 0, *this) ; // copy ouselves across // now add the new row for (int i = 0 ; i < m_NumColumns ; ++i) { result.SetElement(i, m_NumRows, pData[i]) ; } *this = result ; // assign result to us } GdiMatrix operator*(const GdiMatrix &other, double value) { GdiMatrix copy(other) ; // just multiply the elements by the value for (int i = 0 ; i < copy.m_NumColumns ; ++i) { for (int j = 0 ; j < copy.m_NumRows ; ++j) { copy.SetElement(i, j, copy.GetElement(i, j) * value) ; } } return copy ; } GdiMatrix GdiMatrix::GetSquareMatrix() const { GdiMatrix copy(*this) ; copy.MakeSquare() ; return copy ; } void GdiMatrix::MakeSquare() { // make the current matrix square by either stepping in the x or y directions // square to the smallest side int size = m_NumColumns ; if (size > m_NumRows) size = m_NumRows ; GdiMatrix work(size, size) ; // construct result double x_step = m_NumColumns / size ; double y_step = m_NumRows / size ; for (int i = 0 ; i < size ; ++i) { for (int j = 0 ; j < size ; ++j) work.SetElement(i, j, GetElement((int)(i * x_step), (int)(j * y_step))) ; } *this = work ; // copy the result to ourselves } GdiMatrix GdiMatrix::GetNormalised(double min, double max) const { GdiMatrix copy(*this) ; copy.Normalise(min, max) ; return copy ; } void GdiMatrix::Normalise(double min, double max) { // get the lower and upper limit values in the matrix // we use the range to normalise double e_min ; double e_max ; GetNumericRange(e_min, e_max) ; double range = e_max - e_min ; double r_range = max - min ; // required range double value ; for (int i = 0 ; i < m_NumColumns ; ++i) { for (int j = 0 ; j < m_NumRows ; ++j) { value = GetElement(i, j) ; value -= e_min ; // 0 - range value /= range ; value *= r_range ; value += min ; SetElement(i, j, value) ; } } } // gets the lowest and highest values in the matrix void GdiMatrix::GetNumericRange(double &min, double &max) const { double e_min = GetElement(0, 0) ; double e_max = e_min ; double value ; for (int i = 0 ; i < m_NumColumns ; ++i) { for (int j = 0 ; j < m_NumRows ; ++j) { value = GetElement(i, j) ; if (value < e_min) e_min = value ; else if (value > e_max) e_max = value ; } } min = e_min ; max = e_max ; } double GdiMatrix::SumColumn(int column) const { ASSERT(column >= 0) ; // bad column ASSERT(column < m_NumColumns) ; // bad column double sum = 0.0 ; for (int i = 0 ; i < m_NumRows ; ++i) sum += GetElement(column, i) ; return sum ; } double GdiMatrix::SumRow(int row) const { ASSERT(row >= 0) ; // bad row ASSERT(row < m_NumRows) ; // bad row double sum = 0.0 ; for (int i = 0 ; i < m_NumColumns ; ++i) sum += GetElement(i, row) ; return sum ; } double GdiMatrix::SumColumnSquared(int column) const { double value = SumColumn(column) ; return (value * value) ; } double GdiMatrix::SumRowSquared(int row) const { double value = SumRow(row) ; return (value * value) ; } // returns the minimum value in a row of the matrix double GdiMatrix::GetRowMin(int row) const { ASSERT(row >= 0) ; ASSERT(row < m_NumRows) ; double value = GetElement(0, row) ; for (int i = 1 ; i < m_NumColumns ; ++i) { if (GetElement(i, row) < value) value = GetElement(i, row) ; } return value ; } // returns the maximum value in a row of the matrix double GdiMatrix::GetRowMax(int row) const { ASSERT(row >= 0) ; ASSERT(row < m_NumRows) ; double value = GetElement(0, row) ; for (int i = 1 ; i < m_NumColumns ; ++i) { if (GetElement(i, row) > value) value = GetElement(i, row) ; } return value ; } // returns the minimum value in a column of the matrix double GdiMatrix::GetColumnMin(int column) const { ASSERT(column >= 0) ; ASSERT(column < m_NumColumns) ; double value = GetElement(column, 0) ; for (int i = 1 ; i < m_NumRows ; ++i) { if (GetElement(column, i) < value) value = GetElement(column, i) ; } return value ; } // returns the maximum value in a column of the matrix double GdiMatrix::GetColumnMax(int column) const { ASSERT(column >= 0) ; ASSERT(column < m_NumColumns) ; double value = GetElement(column, 0) ; for (int i = 1 ; i < m_NumRows ; ++i) { if (GetElement(column, i) > value) value = GetElement(column, i) ; } return value ; } // returns a row as , separated values CString GdiMatrix::GetRowAsText(int row) const { ASSERT(row >= 0) ; // bad row ASSERT(row < m_NumRows) ; // bad row CString text ; CString token ; for (int i = 0 ; i < m_NumColumns ; ++i) { if (i > 0) text += "," ; token.Format("%e", GetElement(i, row)) ; text += token ; } return text ; } // read a line of text from the current file CString GdiMatrix::ReadLine(CFile& file) { CString line("") ; char ch[3] ; DWORD hBytesRead = 0 ; while (true) { hBytesRead = file.Read(&ch[0], 2) ; if (hBytesRead == 2) file.Seek(-1L, CFile::current) ; if (hBytesRead == 0) break ; // end of file if (ch[0] == '\n') { if (ch[1] == '\r') file.Seek(1L, CFile::current) ; break ; } if (ch[0] == '\r') { ch[0] = '\n' ; if (ch[1] == '\n') file.Seek(1L, CFile::current) ; break ; } if (ch[0] != '\015') { // ignore LF characters ch[1] = '\0' ; line += ch ; } } return line ; } int GdiMatrix::GetStringToken(CString source, CString &destination, int start, char ch) { ASSERT(start >= 0) ; // at the end of the source string ? if (start >= source.GetLength()) { destination = "" ; // no token available return source.GetLength() ; // return @ end of string } // skip past any termination characters at the start position while (start < source.GetLength()) { if (ch == source[start]) start++ ; else break ; } // find the next occurance of the terminating character int pos = source.Find(ch, start) ; // find termination character if (pos < 0) { // terminator not found, just return the remainder of the string destination = source.Right(source.GetLength() - start) ; return source.GetLength() ; } // found terminator, get sub string destination = source.Mid(start, pos - start) ; return pos ; }
Fazit.. Du musst aus allen Verfahren das beste für Dich nutzen, keines der
Maßnahmen erreicht ein ahha Erlebnis.Ich Rechteck-Forscher habe festgestellt, das es keine Kreise gibt. Das ist das eigentlich Problem ^^
Viel Spass
Karsten aus Preußen (www.FlexxVision.de)
-
Achromat schrieb:
Alle Punkte...
Schöne Übersicht mit welchen Ansätzen man das Problem angehen kann.
Auch den Codestil passt zum Alter des Threads, den den du nach mehr als 13 Jahren so unsanft wieder ins Leben zurückgerissen hast (SCNR)Finnegan
P.S.: Erste Aussage ist aufrichtig gemeint. Interessante Ideensammlung!
-
Hi,
ja das ist richtig, denn wieder mal kommt es zur direkten Konfrontation mit diesem leidigen Kreis. Gerade in der Messbilderfassung über Sensoren erhält man schöne Kreise mit auch Fehlerhaften stellen. Und dann geht das Theater wieder von vorne los.. Einige Methoden sind nicht anwendbar andere Methoden schnell zu Rechenlastig.
Ja Oldscool Code , was sonnst.. man kann doch infernalste Vorgänge keiner allgemeinen Lösung zum Fraße reichen, man würde scheitern.
Aber das ist wirklich nur ein wildes Samarium aus Tests und alten Wiedergängern.
Man findet aber die Message in dem Gedünst..
"Mami, Mami, ich will nicht länger im Kreis rumlaufen!"
"Halt´s Maul oder ich nagel deinen anderen Fuß auch noch fest!"