在Open CV和C ++中分割RGB图像并使用斑点检测来检测红色形状

Question

I am using open CV and C++. I need to detect some circular red shaped points on a human face. I am using blob detection but it does not seem to work. I have split the image into red,blue and green but I think I am not getting the parameters right. Can anyone help please? Here is my code http://pastebin.com/aGkUz3SG

Thanks

Answer 1

There are multiple problems:

your blob detector parameters don't fit. .minArea is much too big for the sample image you provided in the comment, since the blobs are very small. .blobColor is a problem since your red isn't perfect, see the image of the red channel I added at bottom. In addition, yellow is combination of red and green and might have a very high red value too. So .filterByColor=false might be better if you don't know the value of your red dots.

Here's the red channel, you will see that the red dots have a lower red value than the yellow dots and values are about the same as the values of the blue dots! In the third image, only one of the red dots is clearly visible to me as a human. On the fourth image, the dot next to the eye isn't really visible. All in all it looks like red channel isn't distinct enough to locate the red dots on human skin.

red channel:

I've tried HSV color space, but the hue value isn't really better, but in saturation channel at least all the dots are a bit visible:

hue:

saturation:

value: