OpenCV計算機視覺處理代碼不起作用

我寫了一個方法來在openCV中爲我正在爲競賽而建的機器人做一些視覺處理。幀和過程對象是mat對象和實例字段，並通過我知道的一個事實工作的另一種方法從usb相機流中獲取。當我用另一種方法顯示框架對象時，如果我已經運行了處理方法，框架就不會顯示，但如果我沒有運行，框架就會顯示。框架墊包含未處理的圖像，通過該圖像繪製標記和輪廓，並且在將標記和輪廓繪製到框架上之前，處理墊用於存儲處理的圖像。 [編輯]：我發現它是由cvtColor方法造成的。這是打印錯誤：（-215）（scn == 3 || scn == 4）& &（depth == CV_8U || depth == CV_32F）。OpenCV計算機視覺處理代碼不起作用

//blurs the image to remove false positives 
    Imgproc.GaussianBlur(frame, processed, new Size(17, 17), 3); 

    //we are going to use HSV, not BGR for better filtration 
    Imgproc.cvtColor(processed, processed, Imgproc.COLOR_BGR2HSV); 

    //create scalars to hold high and low thresholds if using BGR 
    /*Scalar lowRange = new Scalar(RobotMap.lowBlueValue, RobotMap.lowGreenValue, RobotMap.lowRedValue); 
    Scalar highRange = new Scalar(RobotMap.highBlueValue, RobotMap.highGreenValue, RobotMap.highRedValue);*/ 

    //create scalars if using HSV 
    Scalar lowRange = new Scalar(RobotMap.lowHue, RobotMap.lowSat, RobotMap.lowVal); 
    Scalar highRange = new Scalar(RobotMap.highHue, RobotMap.highSat, RobotMap.highVal); 

    //removes everything not in our filter range 
    Core.inRange(processed, lowRange, highRange, processed); 

    //mat used to for some of the contour finding 
    //TODO determine if necessary 
    Mat hierarchy = new Mat(); 

    //create an arraylist to hold the unfiltered contours 
    ArrayList<MatOfPoint> contours = new ArrayList<MatOfPoint>(); 

    //find the contours in our image 
    findContours(processed, contours, hierarchy, RETR_LIST, CHAIN_APPROX_NONE); 

    //list of filtered contours 
    ArrayList<MatOfPoint> filteredContours = new ArrayList<MatOfPoint>(); 

    //list of filtered contours as rect objects 
    ArrayList<Rect> rects = new ArrayList<Rect>(); 

    //put our contours into rectangle objects if they pass our conditions 
    for (MatOfPoint contour : contours) { 
     //bounding rect objects are rectangles whose bounderies encompass all of the contour 
     Rect boundingRect = boundingRect(contour); 
     //check to see if we are a tallish rectangle with a largish area 
     if (boundingRect.height > boundingRect.width && boundingRect.area() > RobotMap.minimumArea) { 
      filteredContours.add(contour); 
      rects.add(boundingRect); 
     } 
    } 

    //draw our contours 
    drawContours(frame, filteredContours, -1, new Scalar(0, 0xFF, 0), FILLED); 
    //figure out how many targets there are 
    numTargets = filteredContours.size(); 

    //draw marker at center of all rects 
    if(rects.size() > 0) 
     Imgproc.drawMarker(frame, center(rects), new Scalar(0xFF, 0, 0xFF)); 

    //draw markers to show info on each rect 
    for (Rect rect : rects) { 
     Imgproc.drawMarker(frame, center(rect), new Scalar(0, 0, 0xFF)); 
     Imgproc.drawMarker(frame, rect.br(), new Scalar(0xFF, 0, 0)); 
     Imgproc.drawMarker(frame, rect.tl(), new Scalar(0xFF, 0, 0)); 
    } 
    if(numTargets > 0) 
     center = center(rects).x;

來源

2017-05-04 Nomaxx117

當源或目標Mat的格式對於給定轉換不正確時出現此錯誤。根據錯誤，該函數正在尋找一個Uint8或Float32的數組。

使用cvtColor「in-place」有時會產生問題。我建議爲HSV圖像創建一個新的墊子。

您還可以檢查this example進行顏色轉換。

來源

2017-05-24 03:56:15 Totoro

謝謝。我發現了這個錯誤。事實證明，在自定義攝像頭服務器的啓動期間，我正在抓取框架，框架還沒有被抓住。如果框架爲空，我會在開始處添加一點檢查。由於這是我的第一篇文章，我如何將其標記爲已解決？ – Nomaxx117

OpenCV計算機視覺處理代碼不起作用

回答

相關問題