Adaptive hybrid bilateral filtering algorithm for infrared image

In view of the mixed noise in the infrared image, the adaptive hybrid bilateral filtering was proposed. The principle of bilateral filter was analyzed. It was put forward that the shortage of bilateral filtering which couldn't filter out strong Gauss noise and impulse noise was inherent. Therefore, the average similarity was defined for bilateral filtering. On the basis of the average similarity, noise points were marked with strong noise points which couldn't be filtered out by bilateral filtering and ordinary noise points which could be filtered out by bilateral filtering. To reduce image fuzzy, only the strong noise points were filtered by median filtering. To keep more edge character, ordinary noise points were filtered by adaptive bilateral filtering. Simulation results show that the hybrid bilateral filtering keeps more edge character, and filters out Gaussian noise and the mixed noise which consist of Gaussian noise and impulse noise effectively. The algorithm is superior to several other methods in subjective evaluation and objective evaluation.