车牌识别之一：车牌定位【实践篇】

在上一篇文章提到车牌定位，是利用opencv sample里面的一个例子，觉得理论上可行，但没有动手操作过。今天有空，修改了一下代码，试了几个图片，发现效果还好。特贴出代码，和各位opencv学习者交流，同时希望有高手指点指点。O(∩_∩)O

废话少说，代码如下：

 

//////////////////////////////////////////////////////////////////////////
//////////////////////////////////////////////////////////////////////////
//车牌定位
//modified by sing
//2010-10-14

//
// The full "Square Detector" program.
// It loads several images subsequentally and tries to find squares in
// each image
//
#ifdef _CH_
#pragma package <opencv>
#endif

#define CV_NO_BACKWARD_COMPATIBILITY

#include "cv.h"
#include "highgui.h"
#include <stdio.h>
#include <math.h>
#include <string.h>

int thresh = 50;
IplImage* img = 0;
IplImage* img0 = 0;
CvMemStorage* storage = 0;
const char* wndname = "Square Detection Demo";

// helper function:
// finds a cosine of angle between vectors
// from pt0->pt1 and from pt0->pt2
double angle( CvPoint* pt1, CvPoint* pt2, CvPoint* pt0 )
{
    double dx1 = pt1->x - pt0->x;
    double dy1 = pt1->y - pt0->y;
    double dx2 = pt2->x - pt0->x;
    double dy2 = pt2->y - pt0->y;
    return (dx1*dx2 + dy1*dy2)/sqrt((dx1*dx1 + dy1*dy1)*(dx2*dx2 + dy2*dy2) + 1e-10);
}

// returns sequence of squares detected on the image.
// the sequence is stored in the specified memory storage
CvSeq* findSquares4( IplImage* img, CvMemStorage* storage )
{
    CvSeq* contours;
    int i, c, l, N = 1;
    CvSize sz = cvSize( img->width & -2, img->height & -2 );    //保证最后一位是偶数,by sing 2010-10-11
    IplImage* timg = cvCloneImage( img ); // make a copy of input image
    IplImage* gray = cvCreateImage( sz, 8, 1 );
    IplImage* pyr = cvCreateImage( cvSize(sz.width/2, sz.height/2), 8, 3 );
    IplImage* tgray;
    CvSeq* result;
    double s, t;
    // create empty sequence that will contain points -
    // 4 points per square (the square's vertices)
    CvSeq* squares = cvCreateSeq( 0, sizeof(CvSeq), sizeof(CvPoint), storage );

    // select the maximum ROI in the image
    // with the width and height divisible by 2
    cvSetImageROI( timg, cvRect( 0, 0, sz.width, sz.height ));

    // down-scale and upscale the image to filter out the noise
    cvPyrDown( timg, pyr, 7 );
    cvPyrUp( pyr, timg, 7 );
    tgray = cvCreateImage( sz, 8, 1 );

    // find squares in every color plane of the image
    for( c = 0; c < 3; c++ )
    {
        // extract the c-th color plane
        cvSetImageCOI( timg, c+1 );

        cvCopy( timg, tgray, 0 );

        // try several threshold levels
        //for( l = 0; l < N; l++ )
        {
            // hack: use Canny instead of zero threshold level.
            // Canny helps to catch squares with gradient shading
            //if( l == 0 )
            //{
            //    // apply Canny. Take the upper threshold from slider
            //    // and set the lower to 0 (which forces edges merging)
            //    cvCanny( tgray, gray, 0, thresh, 5 );
            //    // dilate canny output to remove potential
            //    // holes between edge segments
            //    cvDilate( gray, gray, 0, 1 );
            //}
            //else
            {
                // apply threshold if l!=0:
                //     tgray(x,y) = gray(x,y) < (l+1)*255/N ? 255 : 0
                cvThreshold( tgray, gray, /*(l+1)*255/N*/100, 255, CV_THRESH_BINARY );
            }

            // find contours and store them all as a list
            cvFindContours( gray, storage, &contours, sizeof(CvContour),
                CV_RETR_LIST, CV_CHAIN_APPROX_SIMPLE, cvPoint(0,0) );

            // test each contour
            while( contours )
            {
                // approximate contour with accuracy proportional
                // to the contour perimeter
                result = cvApproxPoly( contours, sizeof(CvContour), storage,
                    CV_POLY_APPROX_DP, cvContourPerimeter(contours)*0.02, 0 );
                // square contours should have 4 vertices after approximation
                // relatively large area (to filter out noisy contours)
                // and be convex.
                // Note: absolute value of an area is used because
                // area may be positive or negative - in accordance with the
                // contour orientation
                if( result->total == 4 &&
                    cvContourArea(result,CV_WHOLE_SEQ,0) > 1000 &&
                    cvContourArea(result,CV_WHOLE_SEQ,0) < (sz.width * sz.height / 4) && 
                    cvCheckContourConvexity(result) )
                {
                    s = 0;

                    for( i = 0; i < 5; i++ )
                    {
                        // find minimum angle between joint
                        // edges (maximum of cosine)
                        if( i >= 2 )
                        {
                            t = fabs(angle(
                                (CvPoint*)cvGetSeqElem( result, i ),
                                (CvPoint*)cvGetSeqElem( result, i-2 ),
                                (CvPoint*)cvGetSeqElem( result, i-1 )));
                            s = s > t ? s : t;
                        }
                    }

                    // if cosines of all angles are small
                    // (all angles are ~90 degree) then write quandrange
                    // vertices to resultant sequence
                    if( s < 0.3 )
                        for( i = 0; i < 4; i++ )
                            cvSeqPush( squares,
                            (CvPoint*)cvGetSeqElem( result, i ));
                }

                // take the next contour
                contours = contours->h_next;
            }
        }
    }

    // release all the temporary images
    cvReleaseImage( &gray );
    cvReleaseImage( &pyr );
    cvReleaseImage( &tgray );
    cvReleaseImage( &timg );

    return squares;
}

//获取最大矩形区域 write by sing 2010-10-14
CvRect GetBoundingRect(CvPoint* points, int count = 4)
{
    unsigned int minX = -1, minY = -1;
    unsigned int maxX = 0, maxY = 0;

    for (int i = 0; i < count; i++) {
        if (minX > points[i].x) {
            minX = points[i].x;
        }
        if (minY > points[i].y) {
            minY = points[i].y;
        }
        if (maxX < points[i].x) {
            maxX = points[i].x;
        }
        if (maxY < points[i].y) {
            maxY = points[i].y;
        }
    }

    CvRect rc = cvRect(minX, minY, maxX - minX, maxY - minY);

    return rc;
}


// the function draws all the squares in the image
void drawSquares( IplImage* img, CvSeq* squares )
{
    CvSeqReader reader;
    IplImage* cpy = cvCloneImage( img );
    int i;

    // initialize reader of the sequence
    cvStartReadSeq( squares, &reader, 0 );

    // read 4 sequence elements at a time (all vertices of a square)
    for( i = 0; i < squares->total; i += 4 )
    {
        CvPoint pt[4], *rect = pt;
        int count = 4;

        // read 4 vertices
        CV_READ_SEQ_ELEM( pt[0], reader );
        CV_READ_SEQ_ELEM( pt[1], reader );
        CV_READ_SEQ_ELEM( pt[2], reader );
        CV_READ_SEQ_ELEM( pt[3], reader );

        // draw the square as a closed polyline
        cvPolyLine( cpy, &rect, &count, 1, 1, CV_RGB(0,255,0), 1, CV_AA, 0 );

        //截取图像并判断是否为车牌
        CvRect rc = GetBoundingRect(pt, count);
        //printf("[%d, %d, %d, %d]/n", rc.x, rc.y, rc.width, rc.height);
        IplImage* img2 = cvCreateImage(cvSize(rc.width, rc.height), IPL_DEPTH_8U, 3);
        cvZero(img2);
        cvSetImageROI(img, rc);
        cvCopyImage(img, img2);
        cvResetImageROI(img);
       
        cvNamedWindow("tmp", CV_WINDOW_AUTOSIZE);
        cvShowImage("tmp", img2);
        cvReleaseImage(&img2);
        cvWaitKey(0);
    }

    // show the resultant image
    cvShowImage( wndname, cpy );
    cvReleaseImage( &cpy );
}


char* names[] = {
    "1.bmp", "2.bmp", "3.bmp",
    "4.bmp", "5.bmp", 0
};

int main(int argc, char** argv)
{
    int i, c;
    // create memory storage that will contain all the dynamic data
    storage = cvCreateMemStorage(0);

    for( i = 0; names[i] != 0; i++ )
    {
        // load i-th image
        img0 = cvLoadImage( names[i], 1 );
        if( !img0 )
        {
            printf("Couldn't load %s/n", names[i] );
            continue;
        }
        img = cvCloneImage( img0 );

        // create window and a trackbar (slider) with parent "image" and set callback
        // (the slider regulates upper threshold, passed to Canny edge detector)
        cvNamedWindow( wndname, 1 );

        // find and draw the squares
        drawSquares( img, findSquares4( img, storage ) );

        // wait for key.
        // Also the function cvWaitKey takes care of event processing
        c = cvWaitKey(0);
        // release both images
        cvReleaseImage( &img );
        cvReleaseImage( &img0 );
        // clear memory storage - reset free space position
        cvClearMemStorage( storage );
        if( (char)c == 27 )
            break;
    }

    cvDestroyWindow( wndname );

    return 0;
}