Lip segmentation algorithm based on bi-color space

Lip region segmentation is an important premise of the lip related biometrics. In real-world applications, the segmentation algorithm pays more attention to efficiency and accuracy, which are often susceptible to different lighting conditions. In this paper, based on lots of experiments, a new lip segmentation algorithm with bi-color space is proposed and a lip-color cascaded filter is designed. Furthermore, filtering non-lip-color pixels is realized based on the distribution characteristic of lip-color pixels. The experiments show that the proposed approach does obtain very promising performance and better segmentation integrity in different lighting environments.