Hi,
Im curious if anyone else is having this problem. First I blur the image using fcvFilterGaussian3x3u8 then I run fcvFilterCanny3x3u8 on the blurred image. If there are edges present this works well with appropriate thresholds set. However if I point the camera to a place with no edge (e.g. table top) then the edge image is filled with noise. I have tried adjusting both blurs and thresholds and the noisy image still results. My code is very simple and the important parts are as follows:
fcvFilterGaussian3x3u8(int_img, width,height,blurred_XY1,1);
fcvFilterCanny3x3u8(blurred_XY, width,height,Edge_img,90,100);
The rest is just getting an intensity image and outputting the image. Am I missing something in using these functions? Like I said when edges are present, like an object on the table, the edge image is good, but if it the table top is all that is in the image I get a very noisy edge image.
Thanks,
Chris
I think this is an inherent feature of Canny edge detection algorithm where the quality of edges is dependent on the threshold value/s. Did you try setting the two threshold values to a high number? My understanding is increasing the threshold values will reduce the noise but there is a chance that you will lose some 'real' edges as well.