An image sequence segmentation method using gradient orientation information

This paper describes a method for segmenting image sequences to moving objects and stationary background. The segmentation is based on gradient orientation information (GOI) that is robust to irregular lighting conditions. The GOI can be easily obtained and processed in the form of unit gradient vectors. Simulation results show that the proposed method works better than a traditional image-intensity based background subtraction approach with temporal Median filtering under a constant lighting condition. More importantly, the results also reveal that, unlike the traditional approach, the proposed method is remarkably robust to a sudden change of image intensities.