Smoke Removal Method of Industrial Images Based on Dark Channel Prior Approach and Second-Generation Wavelets

In the field of digital image and computer vision, haze and smoke removal (dehazing) is one pf a popular scientific arena where it is being studied by an ample number of computer scientists. However, conventional joint dehazing and noise removal techniques use traditional hand-crafted approaches in order to achieve clean image by apply suitable restoration processes. In this study, a smoke removal technique, which designed according to dark channel, prior DCP is proposed. Theoretically, traditional DCP utilized the atmospheric light in the entire image as a global constant and as a result, in our study, the suggested DCP is further enhanced by using filter of second-generation wavelets (SGWs) and therefore, more powerful model is achieved. Consequently, the SGWs filtering is utilized due to its strength in custom designing for complex domains and irregular sampling in order to improve the solution of atmospheric light. The proposed algorithm is evaluated against several state of the art smoke removal methods. Extensive experimentation on multiple datasets demonstrates that our method exhibits better dehazing performance and generalizability than other image dehazing approaches in terms of peak signal-to-noise ratio PSNR, the structural similarity index SSIM and the average gradient results. Moreover, the execution efficiency of the proposed algorithm can reduce considerably the processing time in comparison with the state-of-the-art- smoke removal techniques.