FPGA Implementation of Haze Removal Technique Based on Dark Channel Prior

Image dehazing is a much innovative and growing technology in applications of computer vision. Haze removal technique in FPGA using Nexys 4 DDR is implemented in this paper. The dark channel prior (DCP) has been an efficient dehazing technique. However, DCP can induce inaccurate approximation of transmission which results in colour distortion and halo effects in the brighter regions of an image. The proposed algorithm is implemented for haze removal of image to avoid the colour distortion of haze in bright and in nonbright areas with less complexity. The algorithm that is proposed in this paper is compared with the DCP method and the Tarel algorithm using Structural Similarity index (SSIM). The results show that the proposed algorithm removes haze effectively in both bright and non-bright areas of an image and the implementation in FPGA is done with less computational complexity. Implementation of dehazing in FPGA can be used in many applications of computer vision such as surveillance, military and transportation areas.