Outdoor Light Problem In Segmentation
Hi all, I have a problem in my project called by agriculture cv. I have been using PS 3 Eye as a frame grabber. Its sensor is ov534 and it is not able to lighting condition. When i calculate histogram of frame, it gives me false detections in outdoor light conditions. My method is back projection of hue channel after histogram calculating. So how can i solve this problem? Changing camera or color reducing via normalized histogram?
plant's histogram.
Entire image's histogram.
Back Projection Image.
my working on static image.
On static image, everything is normal but in real time i can not get same results because of camera's motion blur and color changing. In outdoor conditions image shines completely so i only can see red vertical stick on entire image's histogram.
Thanks in advance. Sorry for my bad English.
Can you post images of the non-static image and its resulting Back Projection Image? It is true that camera quality will affect your algorithm. The overexposure is particularly bad, cameras (cheap) will automatically try to adjust the amount of captured light and you can't control it. Many cameras do a poor job at it and will severely affect most of cv algorithms, specially one based on color like yours.
Non-static image shines completely. Its histogram is just red. I have been looking for a new camera. But anyway i try to tune camera parameters.