A camera equipped with an optical fish-eye lens has more widely degree ofview (DOV), such a CCTV system will be used to reduce number of cameras. However, due to their low and non-uniform resolutions from transformation into de-warped images, these images are only able to provide moderate quality in traffic surveillance. Especially, some near boundary images of convex images are more plentiful distorted compare with images in the center of convex images. In this paper, the authors use two methods to overcome this distortion, the one is image segmentation by nonlinear diffusion equation and the other is deformation for some segmented area, when the authors use the images obtained from box camera housing mounting fisheye lens. Also, with digital images, we can get de-blur images in real time as current image sequences. To analysis the experimental result, we calculate PSNR toget 5-8 decibel value better than current method.
Abstract