Opencv 2.4.12_2 calcOpticalFlowPyrLK() 错误

Posted

技术标签:

【中文标题】Opencv 2.4.12_2 calcOpticalFlowPyrLK() 错误【英文标题】:Opencv 2.4.12_2 calcOpticalFlowPyrLK() Error 【发布时间】:2016-04-30 13:13:20 【问题描述】:

我正在尝试使用 OpenCV 的 calcOpticalFlowPyrLK() 函数,但我似乎无法克服这个错误:

OpenCV 错误:断言失败 (mtype == type0 || (CV_MAT_CN(mtype) == CV_MAT_CN(type0) && ((1

我检查了输入参数,两个图像都是: 尺寸:[1280 x 720] 深度:(CV_8U)

代码如下:

Mat frame;
Mat back;
Mat fore;
Mat temp;
Mat prevImage;
Mat currImage;
// for floodfill
Point seed = Point(0,0);
VideoCapture cap("./Sequence2/Sequence_03_%03d.jpg");
// Background subtraction
BackgroundSubtractorMOG2 bg;
bg.set("nmixtures",5);
bg.set("detectShadows", true);
bg.set("fTau", 0.5);
//
bool foundpoints = false;
vector<uchar> status, err;
// Blob Detector
SimpleBlobDetector::Params params;
//params.filterByColor = true;
//params.blobColor = 255; // use if bitwise not statement used
// // Filter by Area.
params.filterByArea = true; //size 
params.minArea = 25;
params.maxArea = 300;
params.filterByCircularity = true; // circle or not
params.minCircularity = 0.15;
params.filterByConvexity = true; // closed or not
params.minConvexity = 0.92;
params.filterByInertia = true; // elongated or not
params.minInertiaRatio = 0.40;
SimpleBlobDetector detector(params);
//
std::vector<std::vector<Point> > contours;
std::vector<KeyPoint> keypoints;
std::vector<Point2f> pKeypoints, prevKeypoints;
namedWindow("Video");
//namedWindow("Background");

for (int i = 0; i < 623; i++) 
    file << "./Output3/image" << i << ".jpg";
    cap >> frame;
    // CONVERT TO GRAY
    cvtColor(frame, temp, CV_RGB2GRAY);
    //adaptiveThreshold(frame,temp,1,ADAPTIVE_THRESH_MEAN_C,THRESH_BINARY,3,1);
    bg.operator ()(temp,fore);
    //bg.getBackgroundImage(back);

    // THREHOLD THE IMAGE
    threshold(fore,fore,80,150,THRESH_TOZERO);
    erode(fore,fore,cv::Mat());
    dilate(fore,fore,cv::Mat());
    findContours(fore,contours,CV_RETR_EXTERNAL,CV_CHAIN_APPROX_NONE);
    // REMOVE SOME NOISE
    cv::floodFill(fore, seed,(255));
    erode(fore,fore,cv::Mat());
    dilate(fore,fore,cv::Mat());
    //bitwise_not(fore,fore);
    erode(fore,fore,cv::Mat());
    dilate(fore,fore,cv::Mat());
    detector.detect(fore, keypoints);
    //std::cout << keypoints.size() << "\n";
    if (keypoints.size() > 0)
        if (foundpoints == false)
            foundpoints = true;
        else
            cv::Size winsize = fore.size();
            KeyPoint::convert(keypoints,pKeypoints);
           // std::cout << pKeypoints.size() << " , " << prevKeypoints.size() << " , " << prevImage.depth() << " , "<< fore.depth();
            calcOpticalFlowPyrLK(prevImage,fore,prevKeypoints,pKeypoints,status,err,winsize);
            prevImage.pop_back();
        
        prevImage.push_back(fore);
        KeyPoint::convert(keypoints,prevKeypoints);
    
    imshow("Frame",fore); // frame
    // imshow("Background",back);
    if(cv::waitKey(30) >= 0) break;
    imwrite(file.str(), fore);
    file.str("");

【问题讨论】:

【参考方案1】:

vector&lt;uchar&gt; err 更改为Mat err

【讨论】:

以上是关于Opencv 2.4.12_2 calcOpticalFlowPyrLK() 错误的主要内容,如果未能解决你的问题,请参考以下文章

使用 Visual Studio 2015 编译 OpenCV 2.4.12 时出错

OpenCV 中的 KLT 跟踪器无法与 Python 一起正常工作

ubuntu下装opencv(转)

使用 OpenCV C++ 查找已知对象

opencv3日记_开发环境设置(vs2013+opencv3.2)

OPENCV学习笔记2-8_Mat_类