In this paper, phase congruency (PC) features for different down-sampled and Gaussian blurred images are presented and analyzed. The detailed textures in the background can be suppressed in the PC maps for the down-sampled and blurring images. As the down-sampling ratio increases, the detailed edges and complex textures in the background can be effectively suppressed. For the blurring images, the PC metric shows a good robustness for the highly down-sampled images. Experimental results show that the proposed down-sampling method can effectively suppress the background textures than the traditional PC approach for edge detection for both the normal and blurring images, which can be very useful for the foreground extraction and background modeling problems.
|Title of host publication||International Conference on Wavelet Analysis and Pattern Recognition|
|Publication status||Published - 2015|
- Background modeling
- Burring images
- Edge detection
- Foreground extraction
- Phase congruency