An Object Region Extraction Approach Based on Interest Pixel Detection and Inner Filling Strategy

In object detection based on computer vision, object extraction is very important. A method and apparatus for extracting interest region from an image is proposed in this paper. First, the approach locate all “interest” pixels within the image by extraction function, which compares the brightness of current pixel with the mean of neighboring pixels and set a threshold to determine whether the pixel is “interest” pixels or not, the “non-interest” pixels neighbored each other are grouped to “non-interest” regions, which are background region or small region enclosed by object region. After that, all the “interest” pixels neighbored each other are grouped to a “interest” region as a object region. Last, the extracted objects have some small “hole” because the object contains small smooth regions, the “hole” region enclosed by “interest” region are filled by simple filling approach. The experiment result is presented in the last part of the paper and the proposed approach is feasible and valid.