如何裁剪由OCT扫描的光学图像...，以便仅可见视网膜层

Question

我正在研究图像分类问题，现在尝试裁剪由OCT扫描的光学图像，以仅提取多个图像的视网膜层（大约85000张图像）。所有图像都是不同的（附有少量参考）

请解决我的代码以应用所有可以裁剪感兴趣区域的图像。

我试图找到轮廓并裁剪图像..但是无法这样做，因为图像的背景不是黑色，并且也有像素 （注意：图像不是灰度的）

我的代码：

    new_filepath = "/home/lebenpc/magick_octid"
    for new_name in os.listdir(new_filepath):
        if "." not in new_name:
            continue
        end = new_name.split(".")[1]
        if end not in ["jpeg"]:
            continue

        try:
            # Attempt to open an image file
            image= cv2.imread(os.path.join(new_filepath,new_name))
            gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
            blur = cv2.GaussianBlur(gray, (7,7), 0)
      # threshold the image, then perform a series of erosions +
      # dilations to remove any small regions of noise
            _, thresh = cv2.threshold(blur, 40, 255, cv2.THRESH_BINARY)
         # find contours in thresholded image, then grab the largest
         # one
            contours, _ = cv2.findContours(thresh, cv2.RETR_CCOMP, cv2.CHAIN_APPROX_TC89_KCOS)
            c = max(contours, key = cv2.contourArea)
            x,y,w,h = cv2.boundingRect(c)
            crop =image[y:y+h,x:x+w]
            name,extension = os.path.splitext(new_name)
            cv2.imwrite(os.path.join("/home/lebenpc/abc", name + '.jpeg'),crop)
        except IOError:
                # Report error, and then skip to the next argument
                print ("Problem opening",new_filepath,":")

我期待可以应用于所有图像的代码