IR and visible images registration method based on cross cumulative residual entropy

This paper presents a method which combines with Bilateral Filter and cross cumulative residual entropy. It will be applied to infrared and visible registration. In this algorithm, firstly, according to infrared image and optical image characteristics, we put forward edge extraction algorithm based on the Bilateral Filter. Secondly, we use Cross Cumulative Residual Entropy (CCRE) as the similarity measure to match the reference images and transformed images effectively. Finally, we introduce the idea of calibration to reduce operation time. Bilateral filter can reduce noise and protect edge, and cross cumulative residual entropy uses cumulative distribution function instead of probability density function to overcome the noise on the local minima. The experiment proved that registration is effective.