Shulou(Shulou.com)06/03 Report--

How to use OpenCV in C++ to achieve gesture recognition, many novices are not very clear about this, in order to help you solve this problem, the following editor will explain in detail for you, people with this need can come to learn, I hope you can get something.

First of all, I will give you an effect picture:

The source code is below

Use the RGB value to segment the hand area, that is, the GB value of the hand will be different from the background

Or use edge detection

Or

Background subtraction.

I use the background subtraction model here. OpenCV provides us with different background subtraction models, and the function of codebook is to calibrate some frames accurately over a period of time. For all the images it acquires; it calculates the average and deviation of each pixel and specifies the box accordingly.

In the foreground it is like a black-and-white image, only the hand is white

Use Convex Hull to find your fingertips. Convex hull is basically a convex set that surrounds the area of the hand.

The red line of the encircling hand is convex hull. Basically it is a bulge; if we take any two points in the red area and connect them together to form a line, then the line is completely within the set.

The yellow dot is a defect point, and there will be many such defect points, that is, every valley has a defect point. Now we can calculate the number of fingers unfolded according to the number of defect points.

It's probably just

Hand region extraction is completed by background subtraction.

For the tip, the depth point convexity defect.

The main code for extracting contours and detecting bumps is in the function.

Invalid detection (IplImage* img_8uc1,IplImage* img_8uc3)

Put the camera in front of a stable background; run the code and wait for a while. After the calibration is complete. You will see an image of the connection component showing some interference. Put your hand in the camera view.

There's nothing to say. It's easier to understand if you look at the code directly.

Core code

Int main (int argc, char** argv) {const char* filename = 0; IplImage* rawImage = 0CodeBookCC = 0; CvCapture* capture = 0; int c, n, nframes = 0; int nframesToLearnBG = 300; model = cvCreateBGCodeBookModel (); model- > modMin [0] = 3; model- > modMin [1] = model- > modMin [2] = 3; model- > modMax [0] = 10 Model- > modMax [1] = model- > modMax [2] = 10; model- > cbBounds [0] = model- > cbBounds [1] = model- > cbBounds [2] = 10; bool pause = false; bool singlestep = false; for (n = 1; n

< argc; n++ ) { static const char* nframesOpt = "--nframes="; if( strncmp(argv[n], nframesOpt, strlen(nframesOpt))==0 ) { if( sscanf(argv[n] + strlen(nframesOpt), "%d", &nframesToLearnBG) == 0 ) { help(); return -1; } } else filename = argv[n]; } if( !filename ) { printf("Capture from camera\n"); capture = cvCaptureFromCAM( 0 ); } else { printf("Capture from file %s\n",filename); capture = cvCreateFileCapture( filename ); } if( !capture ) { printf( "Can not initialize video capturing\n\n" ); help(); return -1; } for(;;) { if( !pause ) { rawImage = cvQueryFrame( capture ); ++nframes; if(!rawImage) break; } if( singlestep ) pause = true; if( nframes == 1 && rawImage ) { // CODEBOOK METHOD ALLOCATION yuvImage = cvCloneImage(rawImage); ImaskCodeBook = cvCreateImage( cvGetSize(rawImage), IPL_DEPTH_8U, 1 ); ImaskCodeBookCC = cvCreateImage( cvGetSize(rawImage), IPL_DEPTH_8U, 1 ); cvSet(ImaskCodeBook,cvScalar(255)); cvNamedWindow( "Raw", 1 ); cvNamedWindow( "ForegroundCodeBook",1); cvNamedWindow( "CodeBook_ConnectComp",1); } if( rawImage ) { cvCvtColor( rawImage, yuvImage, CV_BGR2YCrCb ); if( !pause && nframes-1 < nframesToLearnBG ) cvBGCodeBookUpdate( model, yuvImage ); if( nframes-1 == nframesToLearnBG ) cvBGCodeBookClearStale( model, model->

If (nframes-1 > = nframesToLearnBG) {cvBGCodeBookDiff (model, yuvImage, ImaskCodeBook); centers if desired cvCopy (ImaskCodeBook,ImaskCodeBookCC); cvSegmentFGMask (ImaskCodeBookCC); cvShowImage ("CodeBook_ConnectComp", ImaskCodeBookCC) Detect (ImaskCodeBookCC,rawImage);} cvShowImage ("Raw", rawImage); cvShowImage ("ForegroundCodeBook", ImaskCodeBook);} c = cvWaitKey (10) & 0xFF; c = tolower (c); if (c = 27 | c ='q') break Switch (c) {case'hacks: help (); break; case'pads: pause =! pause; break; case's questions: singlestep =! singlestep; pause = false; break; case'rpm: pause = false Singlestep = false; break; case'': cvBGCodeBookClearStale (model, 0); nframes = 0; break Case 'yearly: case' 0: case 'upland: case' 1: case 'vault: case' 2: case'a: case'3: case'bund: ch [0] = c = ='y' | | c = ='0' | | c = ='a' | | c = ='3' Ch [1] = c = ='u' | | c = ='1' | | c = ='a' | | c = ='3' | | c = = 'baked; ch [2] = c = =' v' | | c = ='2' | | c = ='3' | c = = 'baked; printf ("CodeBook YUV Channels active:% d,% d,% d\ n", ch [0], ch [1], ch [2]) Break; case'ifying: case'odyne: case'knight: case'lforth: {uchar* ptr = c = ='i' | | c = ='o'? Model- > modMax: model- > modMin; for (n = 0; n

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.