In recent years, image defogging algorithm using dark channel prior has always been a research hot spot in the field of haze removal. Scholars at home and abroad are also put forward some new methods, but most researchers are dedicated to improving the quality of image, the real time of defogging algorithm has not been very ideal. In practice, most outdoor vision systems require algorithm has good real-time performance. Therefore, research on real-time defogging algorithm has important theoretical significance and practical application value.To avoid color distortion of the sky region and the low efficiency of the original algorithm, an improved algorithm is proposed for single image haze removal using dark channel prior. The proposed method sets the minimum threshold of transmission by judging the hazy image with sky or no, to fix color distortion of the sky region. Considering that most of the computation time is consumed in the step of estimating the transmission in the original algorithm, the new algorithm firstly estimates the transmission of a hazy image after down-sampling, then obtains the final transmission by interpolation. The suitable selection of patch size and scaling factor results that the complexity of the algorithm reduces greatly with just a little degradation of defogging image. The result is To evaluating the performance of the proposed algorithm, image quality and real time of different defogging algorithms are compared and analyzed. Image quality evaluation can be divided into subjective evaluation and objective evaluation. Blind contrast enhancement assessment by gradient ratio at visible edges and no-reference quality assessment method are chosen as objective evaluation. The experimental results show that the proposed algorithm takes about 8ms for a 600 X 400 image on a 2.6 GHz i5 processor; When compared with Kim algorithm, the proposed algorithm improves the new visible edges of defogging image by more than 20%, and improves the speed by more than 50%.The proposed method is a simple, efficient and fast image defogging algorithm. It can be applied to real-time outdoor vision systems, such as vehicle detection in intelligent transportation, video monitoring, etc. |