|
Cool, I have also found
<a href=>"http://www.codeproject.com/Articles/38447/Generic-List-C">Generic List (C#)</a>
Looks like that is my afternoon sorted!
Glenn
|
|
|
|
|
It can insert multiple values or characters in your program. Array of string is the best way to insert multiple character to take at a time with the help of strings. Most of string start from zero. Three kind of strings are there : one pair string,two pair string and multiple pair string.
|
|
|
|
|
Are you really sure you meant to reply to me? The OP doesn't get notified if you post an answer to my posts.
|
|
|
|
|
Thanks for the Help! I now have it getting the data tomorrow I can work on it further,but now have the bones.
Glenn
|
|
|
|
|
You're welcome. I look forward to finding out how you get on.
|
|
|
|
|
You are not the only one, it would easier if I had more direction than oh it work like X when X is not fully tested and the board you are writing the code for does not exist yet, but to keep the powers that be happy (Gannt charts running) "The Windows Interface Code Must Be Started:". Will let you know!
Glenn
(Dilbert anyone??)
|
|
|
|
|
Yes a List, or a Queue if you want to have one thread adding values and another processing them. 
glennPattonWork wrote: strings rather than ints
If they are ints that have been stringified, then why not Parse them and then add them to the collection?
|
|
|
|
|
Good, These are strings, I meant I had used an integer array to store the data as it came in. I was thinking that as Strings are bigger in size I didn't want to slow everything down by trying to handle a load of them as an array, (seemed to me like having one huge box containing smaller boxes rather unwieldy!)
Glenn
|
|
|
|
|
The alternative, as noted in this part of the thread, is to use a Queue so you could remove items when you are finished processing them - this helps to keep things down to a more managable size.
|
|
|
|
|
While text (strings) may be a good way to transfer information between systems, it is not the recommended way to store data in memory. So my approach would be slightly different, I'd have:
- a textual representation of the data, as an example X123Y456E would be a point with X and Y coordinates (E stands for a terminator, think newline character);
- a class that holds an instance of data, in the example:
public class MyPoint {
public int X;
public int Y;
}
BTW: I prefer a class over a struct to avoid all kinds of problems including boxing/unboxing.
- some container that holds all the data, this could be List<MyPoint> myCollectionOfPoints in the example.
- a method to receive textual data, convert it on the spot, and store it; it might look like:
public void ReceiveMyPoint() {
string line=mySerialPort.ReadLine();
... here comes input validation and parsing, resulting in actual values:
int x=...;
int y=...;
myCollectionOfPoints.Add(new MyPoint(x,y));
}
- a thread to consume all incoming data, basically a loop calling ReceiveMyPoint() all the time. I don't use the DataReceived event here, as that does not synchronize with incoming text lines, it fires randomly when "some data" is available, which may well be a partial line of text.
The net result is your interface is textual, the data is in meaningful types (two ints in the example), conversion is not a separate step, and no storage space gets wasted.
Note: you must program defensively when receiving data from another system, as it will go wrong sooner or later, with a bad conenction, some missing characters, a noisy line, whatever. Therefore validation is important; even the simplest checksum or CRC added to each message would prove valuable.
|
|
|
|
|
Great thanks for that, can you recommend a reference for this type of data handling (or really any data handling would be useful) It's just I am tasked with writing an application to interface to a board that does exist (yet!) "it will do it like the X board, you already written the software for that". I'm looking at this as a chance to increase my Windows knowledge!
Glenn
|
|
|
|
|
Sorry, no reference, just my own experience with peripherals and embedded systems.
|
|
|
|
|
Ah well google & MSDN, I did look at something on the code project that confused me! (link earlier)
Glenn
|
|
|
|
|
Thanks for the Help! I now have it getting the data tomorrow I can work on it further,but now have the bones.
Glenn
|
|
|
|
|
using System;
using System.Collections.Generic;
using System.Text;
using System.Drawing;
using System.Collections;
using System.Windows.Forms;
class SIFT
{ static double initDev = 1.6;
static public int OctavesCount = 2;
static public int scalesPerOctave = 4;
static double minDoG = 0.01;
static public double norm;
static double edgeRatio = 10;
public static double[][,] BuildGaussianBlurPyramid(Bitmap image)
{ int ind=1;
int imgCount = scalesPerOctave * OctavesCount;
double deviation = initDev;
double[][,] imgMatrix=new double[imgCount][,];
imgMatrix[0]=new double[image.Width,image.Height];
imgMatrix[0]=ImageProcessing.BitmapToGrayscaleMatrix(image);
for (int i = 0; i < OctavesCount; i++){
for (int j = 1; j < scalesPerOctave; j++){
imgMatrix[ind]=ImageProcessing.GaussianBlurBWMat(imgMatrix[ind-1], deviation);
deviation *= 2;
ind++;
}
if (ind<imgCount-1)
imgMatrix[ind] = ImageProcessing.SubSample(imgMatrix[ind-2]);
deviation /= 3.2;
ind++;
}
return imgMatrix;
}
public static double[][,] BuildDoGPyramid(double[][,] gaussianPyr)
{ int ind = 0;
double[][,] resDoG = new double[(scalesPerOctave - 1) * OctavesCount][,];
for (int i = 0; i < gaussianPyr.GetLength(0) - 1; i += 1){
if ((i + 1) % (scalesPerOctave) == 0){
continue;
}
resDoG[ind] = ImageProcessing.CalculateDoGMatrix(gaussianPyr[i], gaussianPyr[i + 1]);
ind++;
}
return resDoG;
}
public static ArrayList FindKeypoints(double[][,] imgMatrixPyr)
{ ArrayList KpL = new ArrayList();
double[][][,] imgMatrix = new double[2][][,];
norm = 3.0 / (initDev * 2);
int DoGCount = (scalesPerOctave - 1) * OctavesCount;
imgMatrix[0] = new double[3][,];
imgMatrix[1] = new double[3][,];
int pyrL = 0;
int pyrIm = 0;
for (int i = 0; i < DoGCount; i++){
if (i == 3){
pyrL++;
pyrIm = 0;
}
imgMatrix[pyrL][pyrIm] = imgMatrixPyr[i];
pyrIm++;
}
int min = 1, max = 1;
for (int l = 0; l < imgMatrix.GetLength(0); l++){
for (int y = 1; y < imgMatrix[l][1].GetLength(1) - 1; y++)
for (int x = 1; x < imgMatrix[l][1].GetLength(0) - 1; x++){
double point = imgMatrix[l][1][x, y];
if (Math.Abs(point) <= minDoG)
continue;
for (int i = 0; i < imgMatrix[l].GetLength(0); i++){
if (max == 1 & imgMatrix[l][i][x - 1, y - 1] < point & imgMatrix[l][i][x - 1, y] < point &
imgMatrix[l][i][x, y - 1] < point & imgMatrix[l][i][x, y] <= point & imgMatrix[l][i][x + 1, y + 1] < point &
imgMatrix[l][i][x + 1, y] < point & imgMatrix[l][i][x, y + 1] < point & imgMatrix[l][i][x - 1, y + 1] < point &
imgMatrix[l][i][x + 1, y - 1] < point){
max = 1;
}
else max = 0;
if (min == 1 & imgMatrix[l][i][x - 1, y - 1] > point & imgMatrix[l][i][x - 1, y] > point &
imgMatrix[l][i][x, y - 1] > point & imgMatrix[l][i][x, y] >= point & imgMatrix[l][i][x + 1, y + 1] > point &
imgMatrix[l][i][x + 1, y] > point & imgMatrix[l][i][x, y + 1] > point & imgMatrix[l][i][x - 1, y + 1] > point &
imgMatrix[l][i][x + 1, y - 1] > point){
min = 1; }
else min = 0;
}
if (max == 1 || min == 1){
Keypoint kp = new Keypoint();
kp.x = x;
kp.y = y;
kp.value = point;
kp.layer = l;
KpL.Add(kp); }
min = 1; max = 1;
}
}
return KpL;
}
public static ArrayList KeypointLocalization(ArrayList initKpList, double[][,] DoGMatrix, int edgeRatio)
{ Keypoint initKp;
ArrayList retKp = new ArrayList();
double Dxx, Dyy, Dxy, Trace, Det;
double r = ((edgeRatio + 1) * (edgeRatio + 1)) / 2;
for (int i = 0; i < initKpList.Count; i++){
initKp = (Keypoint)initKpList[i];
int l = initKp.layer == 0 ? 1 : 4;
if (initKp.x <= 0 || initKp.y <= 0 ||
initKp.x >= DoGMatrix[l].GetLength(0) - 1
|| initKp.y >= DoGMatrix[l].GetLength(0) - 1)
continue;
Dxx = DoGMatrix[l][initKp.x + 1, initKp.y]
+ DoGMatrix[l][initKp.x - 1, initKp.y]
- 2.0 * DoGMatrix[l][initKp.x, initKp.y];
Dyy = DoGMatrix[l][initKp.x, initKp.y + 1]
+ DoGMatrix[l][initKp.x, initKp.y - 1]
- 2.0 * DoGMatrix[l][initKp.x, initKp.y];
Dxy = 0.25 * ((DoGMatrix[l][initKp.x + 1, initKp.y + 1]
- DoGMatrix[l][initKp.x + 1, initKp.y - 1]) -
(DoGMatrix[l][initKp.x - 1, initKp.y + 1]
- DoGMatrix[l][initKp.x - 1, initKp.y - 1]));
Trace = Dxx + Dyy;
Det = Dxx * Dyy - (Dxy * Dxy);
if ((Trace * Trace) / Det > r)
continue;
else
retKp.Add(initKp);
}
return retKp;
}
public static ArrayList AssignOrientations(ArrayList finKpList, double[][,] blurredMatrix)
{ Keypoint finKp;
ArrayList orKp=new ArrayList();
double binVal = 2 * Math.PI / 36;
for (int i = 0; i < finKpList.Count; i++){
finKp = (Keypoint)finKpList[i];
finKp.scale = initDev*(finKp.layer + 1);
double[] histBin = new double[36];
int kpReg=(int)(1.5*finKp.scale/2+0.5);
double SigmaFact=2*2.25*finKp.scale*finKp.scale;
int layer=finKp.layer==0?1:5;
int regXSt=finKp.x-kpReg<1?1:finKp.x-kpReg;
int regXEnd = finKp.x + kpReg > blurredMatrix[layer].GetLength(0) ? blurredMatrix[layer].GetLength(0) : finKp.x + kpReg;
int regYSt = finKp.y - kpReg < 1 ? 1 : finKp.y - kpReg;
int regYEnd = finKp.y + kpReg > blurredMatrix[layer].GetLength(1) ? blurredMatrix[layer].GetLength(1) : finKp.y + kpReg;
for (int y=regYSt; y<regYEnd; y++)
for (int x=regXSt; x<regXEnd; x++){
double gaussianWight = Math.Exp(-(Math.Pow(x - finKp.x, 2) + Math.Pow(y - finKp.y, 2) / SigmaFact));
double portInHist = (CalMagnitudeAndDirection(x,y, blurredMatrix[layer]).direction + Math.PI) / (2 * Math.PI);
int ind = (int)(portInHist * 36 == 36 ? 0 : portInHist * 36);
histBin[ind]+=CalMagnitudeAndDirection(x,y,blurredMatrix[layer]).magnitude*gaussianWight;
}
double max = 0.0;
int maxBin = 0;
for (int bi = 0; bi < histBin.GetLength(0); ++bi){
if (histBin[bi] > max){
max = histBin[bi];
maxBin = bi; }
}
max = maxBin * binVal - Math.PI;
max=max<-Math.PI?2*Math.PI+max:max;
max = max > Math.PI ? max - 2 * Math.PI : max;
finKp.orientation = max;
orKp.Add(finKp);
}
return orKp;
}
static KpOrientPar CalMagnitudeAndDirection(int x, int y, double[,] blurredMatrix)
{ KpOrientPar kpO=new KpOrientPar();
kpO.magnitude = Math.Sqrt (Math.Pow (blurredMatrix[x + 1, y] - blurredMatrix[x - 1, y], 2.0) +
Math.Pow (blurredMatrix[x, y + 1] - blurredMatrix[x, y - 1], 2.0));
kpO.direction = Math.Atan2(blurredMatrix[x, y + 1] - blurredMatrix[x, y - 1],
blurredMatrix[x + 1, y] - blurredMatrix[x - 1, y]);
return kpO;
}
public static ArrayList CreateDescriptor(ArrayList finKpList, double[][,] blurredMatrix)
{ int kpReg = 16;
Keypoint finKp;
ArrayList descKp=new ArrayList();
double SigmaFact = 2*(kpReg / 2) * (kpReg / 2);
for (int kpi = 0; kpi < finKpList.Count; kpi++){
finKp = (Keypoint)finKpList[kpi];
int regXSt = finKp.x - kpReg < 1 ? 1 : finKp.x - kpReg;
int regYSt = finKp.y - kpReg < 1 ? 1 : finKp.y - kpReg;
int regXEnd = regXSt + 4;
int regYEnd = regXSt + 4;
double[,][] desc = new double[4, 4][];
for (int jj = 0; jj < 4; jj++)
for (int ii = 0; ii < 4; ii++ )
desc[ii,jj] = new double[8];
int layer = finKp.layer == 0 ? 1 : 5;
for (int i = 0; i <= 3; i++)
for (int j =0 ; j <= 3; j++)
for (int xi = regXSt; xi < regXEnd; xi++)
for (int yi = regYSt; yi < regYEnd; yi++){
int x =xi+j*4 >= blurredMatrix[layer].GetLength(0)-1 ? blurredMatrix[layer].GetLength(0)-2 : xi+j*4;
int y = yi+i*4 >= blurredMatrix[layer].GetLength(1)-1 ? blurredMatrix[layer].GetLength(1)-2 : yi+i*4;
double yRot = Math.Sin(-finKp.orientation) * (x - finKp.x) +
Math.Cos(-finKp.orientation) * (y - finKp.y);
double xRot = Math.Cos(-finKp.orientation) * (x - finKp.x) -
Math.Sin(-finKp.orientation) * (y - finKp.y);
double gaussianWight = Math.Exp(-(Math.Pow(xRot, 2) + Math.Pow(yRot, 2) / SigmaFact));
double direction = CalMagnitudeAndDirection(x, y, blurredMatrix[layer]).direction - finKp.orientation;
if (direction < 0)
direction += 2.0 * Math.PI;
else if (direction > 2.0 * Math.PI)
direction -= 2.0 * Math.PI;
double portInHist = direction / (2 * Math.PI);
int ind = (int)(portInHist * 8 == 8 ? 0 : portInHist * 8);
desc[i, j][ind] += CalMagnitudeAndDirection(x, y, blurredMatrix[layer]).magnitude * gaussianWight;
}
finKp.descriptor=desc;
descKp.Add(finKp);
}
return descKp;
}
}
class KpOrientPar
{
public double magnitude;
public double direction;
}
|
|
|
|
|
help me to debug this program ..thanx alot
|
|
|
|
|
You want to know what this code do?
|
|
|
|
|
In the code that calls the method you are getting an exception in, put a breakpoint on the line that calls the method (you can do this by pressing F9). Run your program in debug mode and when the program hits the breakpoint, it will pause execution. At this point, press F11 to step into the method. Check that the parameters that you have passed in are what you would expect. Now, use the step over command (F10) to step over individual code lines, and use step into (F11) to inspect methods that are of interest to you.
Check the values of variables as you progress to determine where the calculation goes wrong (if that's what your problem is), and to check that the right value gets allocated to the right place. This is going to require a lot of patience on your part, so take your time. It also means that you are going to have to be aware of what values you should get at what point.
And that, in a nutshell, is how you debug it.
|
|
|
|
|
Sorry for the inconvenience
please try the program and debug the error then paste it here
thanx
|
|
|
|
|
That's not going to happen. YOU have to debug the program. We're telling you how to do that. It's up to YOU to do the work. We're not getting paid to do your job, YOU are.
|
|
|
|
|
I wish I'd realised that I never had to debug a program again, I could just post the code here and someone would do it for me. How exciting.
|
|
|
|
|
And it's FREE!! Why the Hell haven't we done this before!?
|
|
|
|
|
Did you know that if you put a scarf on a snowball, it stands a really good chance of surving in hades? Your chances of me debugging and working out what the problem is with your code is absolutely nil. Unless you're happy to pay my rates, which I have to warn you are very expensive.
|
|
|
|
|
i have a multi programs in c# that does not execute plz help me,,,i want to know the error in this programs
first:
/* SIFT.cs
* The SIFT algorithm class
* where all the algorithm specific methods
* are implemented
*/
using System;
using System.Collections.Generic;
using System.Text;
using System.Drawing;
using System.Collections;
using System.Windows.Forms;
class SIFT
{ static double initDev = 1.6; // initial standard deviation variable
static public int OctavesCount = 2; // Octaves count for pyramid
static public int scalesPerOctave = 4; // scales(blurred images) count per octave
static double minDoG = 0.01; // minimal threshold of keypoint value
static public double norm; // normalization factor for kp coordinates
static double edgeRatio = 10; // minimal edge ratio for kp localization
public static double[][,] BuildGaussianBlurPyramid(Bitmap image) // method to build the Gaussian pyramid with bitmap input
{ int ind=1; // index variable to help for image indexing in pyramid
int imgCount = scalesPerOctave * OctavesCount; // calculate total image count
double deviation = initDev; // local initial deviation variable
double[][,] imgMatrix=new double[imgCount][,]; // array to store values of each image matrix in the pyramid
imgMatrix[0]=new double[image.Width,image.Height]; // image matrix initial dimension
imgMatrix[0]=ImageProcessing.BitmapToGrayscaleMatrix(image); // create grayscale image matrix from bitmap coloured image
for (int i = 0; i < OctavesCount; i++){
for (int j = 1; j < scalesPerOctave; j++){ // Build Gaussian blur with given image mtrx and deviation
imgMatrix[ind]=ImageProcessing.GaussianBlurBWMat(imgMatrix[ind-1], deviation);
deviation *= 2; // double the previous deviation
ind++; // increment current image index
}
if (ind<imgcount-1) check="" for="" current="" image="" index="" and="" build="" next="" octave
="" imgmatrix[ind]="ImageProcessing.SubSample(imgMatrix[ind-2]);" down-sample="" the="" by="" factor="" of="" 2="" octave=""
="" deviation="" =="" 3.2;="" divide="" to="" start="" from="" twice="" initial
="" ind++;="" increment="" index
="" }
="" return="" imgmatrix;="" resulting="" matrix="" pyramid="" array
="" }="" method="" dog="" pyramid
="" public="" static="" double[][,]="" builddogpyramid(double[][,]="" gaussianpyr)="" {="" int="" ind="0;" value="" controlling="" indxes="" resdog="new" double[(scalesperoctave="" -="" 1)="" *="" octavescount][,];
="" (int="" i="0;" <="" gaussianpyr.getlength(0)="" 1;="" +="1){
" if="" ((i="" %="" (scalesperoctave)="=" 0){="" go="" continue;
="" calculate="" matrix
="" resdog[ind]="ImageProcessing.CalculateDoGMatrix(gaussianPyr[i]," gaussianpyr[i="" 1]);
="" }
="" resdog;="" arraylist="" findkeypoints(double[][,]="" imgmatrixpyr)="" find="" keypoints
="" kpl="new" arraylist();="" create="" store="" double[][][,]="" imgmatrix="new" double[2][][,];="" array="" three="" level="" pyr="" img="" norm="3.0" (initdev="" 2);="" normalization="" factor
="" dogcount="(scalesPerOctave" octavescount;="" count="" total="" number="" images="" imgmatrix[0]="new" double[3][,];="" first="" imgmatrix[1]="new" second="" pyrl="0;" layer="" pyrim="0;" in="" layer
="" dogcount;="" i++){="" (i="=" 3){
="" pyrl++;="" assign="" zero="" dimension="" imgmatrix[pyrl][pyrim]="imgMatrixPyr[i];" copy="" pyr.="" 1="" dim="" dim.
="" pyrim++;="" min="1," max="1;" initialization
="" l="0;" imgmatrix.getlength(0);="" l++){
="" y="1;" imgmatrix[l][1].getlength(1)="" y++)
="" x="1;" imgmatrix[l][1].getlength(0)="" x++){
="" double="" point="imgMatrix[l][1][x," y];="" pick="" one="" central="" (math.abs(point)="" keypoint="" threshold
="" imgmatrix[l].getlength(0);="" i++){
="" maximum="" (max="=" &="" imgmatrix[l][i][x="" 1,="" 1]="" y]="" imgmatrix[l][i][x,="" point){
="" else="" minimum
="" (min="="> point & imgMatrix[l][i][x - 1, y] > point &
imgMatrix[l][i][x, y - 1] > point & imgMatrix[l][i][x, y] >= point & imgMatrix[l][i][x + 1, y + 1] > point &
imgMatrix[l][i][x + 1, y] > point & imgMatrix[l][i][x, y + 1] > point & imgMatrix[l][i][x - 1, y + 1] > point &
imgMatrix[l][i][x + 1, y - 1] > point){
min = 1; }
else min = 0;
}
if (max == 1 || min == 1){ // check if the point is max or min
Keypoint kp = new Keypoint(); // create keypoint object to
kp.x = x; // assign x coordinate of extreme point
kp.y = y; // assign y coordinate of extreme point
kp.value = point; // assign extreme point value
kp.layer = l; // assign layer of extreme point
KpL.Add(kp); } // add keypoint to arraylist
min = 1; max = 1; // reinitialize min and max values
}
}
return KpL; // return keypoint list
}
// method to localize keypoints for the given edge ratio
public static ArrayList KeypointLocalization(ArrayList initKpList, double[][,] DoGMatrix, int edgeRatio)
{ Keypoint initKp; // Keypoint object to store initial keypoints
ArrayList retKp = new ArrayList(); // create Arraylist to store refined keypoints
double Dxx, Dyy, Dxy, Trace, Det; // variables to store elements, trace and Det. of Hessian
double r = ((edgeRatio + 1) * (edgeRatio + 1)) / 2; // calculate the threshold from the edgratio
for (int i = 0; i < initKpList.Count; i++){
initKp = (Keypoint)initKpList[i]; // cast the value from Arraylist object to the Keypoint
int l = initKp.layer == 0 ? 1 : 4; // identify the current layer
if (initKp.x <= 0 || initKp.y <= 0 || // drop keypoints which are on the borders
initKp.x >= DoGMatrix[l].GetLength(0) - 1
|| initKp.y >= DoGMatrix[l].GetLength(0) - 1)
continue;
Dxx = DoGMatrix[l][initKp.x + 1, initKp.y] // calculate the x,x derivative by pixel differences
+ DoGMatrix[l][initKp.x - 1, initKp.y]
- 2.0 * DoGMatrix[l][initKp.x, initKp.y];
Dyy = DoGMatrix[l][initKp.x, initKp.y + 1] //calculate the y,y derivative by pixel differences
+ DoGMatrix[l][initKp.x, initKp.y - 1]
- 2.0 * DoGMatrix[l][initKp.x, initKp.y];
Dxy = 0.25 * ((DoGMatrix[l][initKp.x + 1, initKp.y + 1] // calculate the x,y derivative by pixel differences
- DoGMatrix[l][initKp.x + 1, initKp.y - 1]) -
(DoGMatrix[l][initKp.x - 1, initKp.y + 1]
- DoGMatrix[l][initKp.x - 1, initKp.y - 1]));
Trace = Dxx + Dyy; // calculate the trace of Hessian
Det = Dxx * Dyy - (Dxy * Dxy); // calculate the determinant of Hessian
if ((Trace * Trace) / Det > r) // drop the keypoints which contradict the given ratio
continue;
else
retKp.Add(initKp); // add the remaining keypoints
}
return retKp; // return the resulting Arraylist with localized keypoints
}
// method for orientation assignment
public static ArrayList AssignOrientations(ArrayList finKpList, double[][,] blurredMatrix)
{ Keypoint finKp; // keypoint variable to store localized keyoints
ArrayList orKp=new ArrayList(); // create arraylist to store keypoints with orientations
double binVal = 2 * Math.PI / 36; // calculate the angle value of one histogram bin
for (int i = 0; i < finKpList.Count; i++){
finKp = (Keypoint)finKpList[i]; // cast from array list object to keypoint object
finKp.scale = initDev*(finKp.layer + 1); // caclulate the keypoint scale
double[] histBin = new double[36]; // create the double array as a colour histogram
int kpReg=(int)(1.5*finKp.scale/2+0.5); // calculate the radius of the region around the keypoint
double SigmaFact=2*2.25*finKp.scale*finKp.scale; // caclutate the sigma factor for gaussian weight
int layer=finKp.layer==0?1:5; // identify the current keypoint layer
int regXSt=finKp.x-kpReg<1?1:finKp.x-kpReg; // calculate the x coordinate of the region start and end
int regXEnd = finKp.x + kpReg > blurredMatrix[layer].GetLength(0) ? blurredMatrix[layer].GetLength(0) : finKp.x + kpReg;
int regYSt = finKp.y - kpReg < 1 ? 1 : finKp.y - kpReg; // calculate the x coordinate of the region start and end
int regYEnd = finKp.y + kpReg > blurredMatrix[layer].GetLength(1) ? blurredMatrix[layer].GetLength(1) : finKp.y + kpReg;
for (int y=regYSt; y<regyend; y++)
="" for="" (int="" x="regXSt;" x<regxend;="" x++){="" calculate="" the="" gaussian="" wight=""
="" double="" gaussianwight="Math.Exp(-(Math.Pow(x" -="" finkp.x,="" 2)="" +="" math.pow(y="" finkp.y,="" sigmafact));
="" portion="" of="" direction="" angle="" in="" histogram
="" portinhist="(CalMagnitudeAndDirection(x,y," blurredmatrix[layer]).direction="" math.pi)="" (2="" *="" math.pi);
="" identify="" bin="" index="" current="" keypoint="" int="" ind="(int)(portInHist" 36="=" ?="" 0="" :="" 36);="" add="" magnitude="" weightened="" by="" to="" respective="" bin
="" histbin[ind]+="CalMagnitudeAndDirection(x,y,blurredMatrix[layer]).magnitude*gaussianWight;
" }
="" max="0.0;" initialize="" maximal="" value
="" maxbin="0;" index
="" bi="0;" <="" histbin.getlength(0);="" ++bi){
="" if="" (histbin[bi]=""> max){ // check if the next bin is bigger
max = histBin[bi]; // store new bin as the maximal
maxBin = bi; } // mark the new bin index as the index of the maximal bin
}
max = maxBin * binVal - Math.PI; // change angle range from [0, 2PI] to [-PI,PI]
max=max<-Math.PI?2*Math.PI+max:max; // check if angle is outside [-PI,PI]
max = max > Math.PI ? max - 2 * Math.PI : max; // check if angle is outside [-PI,PI]
finKp.orientation = max; // assign maximal bin direction as keypoint orientation
orKp.Add(finKp); // add keypoint with assign orientation
}
return orKp; // return keypoint list of keypoints with their orientations
}
// method to calculate keypoint magnitude and direction
static KpOrientPar CalMagnitudeAndDirection(int x, int y, double[,] blurredMatrix)
{ KpOrientPar kpO=new KpOrientPar(); // create orientation object for magnitude and direction pair
// calculate the keypoint magnitude by pixel differences
kpO.magnitude = Math.Sqrt (Math.Pow (blurredMatrix[x + 1, y] - blurredMatrix[x - 1, y], 2.0) +
Math.Pow (blurredMatrix[x, y + 1] - blurredMatrix[x, y - 1], 2.0));
// calculate the keypoint direction by pixel differences
kpO.direction = Math.Atan2(blurredMatrix[x, y + 1] - blurredMatrix[x, y - 1],
blurredMatrix[x + 1, y] - blurredMatrix[x - 1, y]);
return kpO; // return orientation object with magnitude and direction
}
public static ArrayList CreateDescriptor(ArrayList finKpList, double[][,] blurredMatrix)
{ int kpReg = 16; // region size around the keypoint
Keypoint finKp; // keypoint object to store input keypoint
ArrayList descKp=new ArrayList(); // arraylist keypoint to store descriptor assigned keypoints
double SigmaFact = 2*(kpReg / 2) * (kpReg / 2); // calculate the sigma factor for gaussian weight
for (int kpi = 0; kpi < finKpList.Count; kpi++){
finKp = (Keypoint)finKpList[kpi]; // cast from arraylist object to keypoint object
int regXSt = finKp.x - kpReg < 1 ? 1 : finKp.x - kpReg; // calculate x coordinate of keypoint region start
int regYSt = finKp.y - kpReg < 1 ? 1 : finKp.y - kpReg; // calculate y coordinate of keypoint region start
int regXEnd = regXSt + 4; // calculate x coordinate of keypoint region end
int regYEnd = regXSt + 4; // calculate y coordinate of keypoint region end
double[,][] desc = new double[4, 4][]; // create double array to store descriptor
for (int jj = 0; jj < 4; jj++)
for (int ii = 0; ii < 4; ii++ )
desc[ii,jj] = new double[8]; // initialize double array of descriptor
int layer = finKp.layer == 0 ? 1 : 5;
for (int i = 0; i <= 3; i++)
for (int j =0 ; j <= 3; j++)
for (int xi = regXSt; xi < regXEnd; xi++)
for (int yi = regYSt; yi < regYEnd; yi++){ // calculate current value of x and y coordinates
int x =xi+j*4 >= blurredMatrix[layer].GetLength(0)-1 ? blurredMatrix[layer].GetLength(0)-2 : xi+j*4;
int y = yi+i*4 >= blurredMatrix[layer].GetLength(1)-1 ? blurredMatrix[layer].GetLength(1)-2 : yi+i*4;
double yRot = Math.Sin(-finKp.orientation) * (x - finKp.x) +
Math.Cos(-finKp.orientation) * (y - finKp.y); // rotate x coordinate by keypoint orientation value
double xRot = Math.Cos(-finKp.orientation) * (x - finKp.x) -
Math.Sin(-finKp.orientation) * (y - finKp.y); // rotate y coordinate by keypoint orientation value
// calculate gaussian weight
double gaussianWight = Math.Exp(-(Math.Pow(xRot, 2) + Math.Pow(yRot, 2) / SigmaFact));
// calculate direction and rotate by keypoint orientation
double direction = CalMagnitudeAndDirection(x, y, blurredMatrix[layer]).direction - finKp.orientation;
if (direction < 0) // check if direction degree is out of [0,2PI] range
direction += 2.0 * Math.PI;
else if (direction > 2.0 * Math.PI)
direction -= 2.0 * Math.PI;
double portInHist = direction / (2 * Math.PI); // calculate the protion of the direction in the histogram
// idetify the index of orientation bin
int ind = (int)(portInHist * 8 == 8 ? 0 : portInHist * 8);
// add magn. weighted by gaussian to the resp.direction bin
desc[i, j][ind] += CalMagnitudeAndDirection(x, y, blurredMatrix[layer]).magnitude * gaussianWight;
}
finKp.descriptor=desc; // assign descriptor to keypoint object
descKp.Add(finKp); // add keypoint to the resulting arraylist object
}
return descKp; // return arraylist of keypoints with descriptors
}
}
class KpOrientPar // helper class to store magnitude and direction
{
public double magnitude; // double variable to store keypoint magnitude
public double direction; // double variable to store keypoint direction
}
|
|
|
|
|
First, when you put code into a forum message, please use the code button and choose the appropriate language. What you do is highlight the code you've inserted, click code and then choose C# from the drop down that appears. As it stands, we cannot even read your code.
Second, what do you mean by "it does not execute"? Do you mean you get compilation errors? Does the program start and then immediately stop? Does it have logic errors?
It's highly unlikely that anyone is going to take your code, drop it into a program and debug it for you. Now's the time for you to learn valuable debugging skills. Set breakpoints. Inspect variables, step through code and do all the other things that you will be expected to do as a professional coder.
|
|
|
|
|
General 
News 
Suggestion 
Question 
Bug 
Answer 
Joke 
Praise 
Rant 
Admin 
Submit an article or tip
