New technique for enhancement of high dynamic range infrared images

A Bilateral Filter (BF) and Multi-scale Retinex (MSR) based infrared image enhancement algorithm is proposed in this paper. It’s known that the MSR algorithm derived from vision theory can achieve dynamic range compression and tonal rendition effectively but suffers from ‘halo’ phenomena caused by the existence of “sharp” edges in infrared images. Research shows that bilateral filter has the property of separating image details from strong edges. Therefore, we process detail components containing strong edges in MSR algorithm with bilateral filter to achieve dynamic range compression, detail enhancement and avoid ‘halo’ artifacts at the same time. The performance of proposed algorithm is then validated by experiments with three real infrared images and compared with other two infrared images enhancement algorithms.