摘要:目的: 基于暗原色先验的单幅图像去雾算法较好地解决了估算雾天场景

摘要:目的: 基于暗原色先验的单幅图像去雾算法较好地解决了估算雾天场景透射率的问题,算法复原的图像清晰颜色自然。但原始算法计算复杂度高、耗时长、无法满足实时性需求。方法: 本文提出了一种基于暗通道先验的改进算法方法,通过改进暗通道图像计算方法,并采用图像引导滤波来优化透射图,使本文提出的方法相比较前人的方法在算法处理速度上有了很大的提高,并且得到的不错的图像视觉效果,同时给出了4种方法的实验结果以及评价分析. 结果:基于导向滤波暗原色方法从主观评价可以看出处理后图像颜色不失真,对比度均衡,细节还原准确,从客观评价均优于其他三种方法,同时处理速度可以满足实时要求 结论: 与各种算法相比,细节增强和颜色保真方面能够取得更好的效果.

結果 (英文) 1:

Abstract: Objective: single image dark channel prior to fog algorithm can effectively solve the estimation problem based on fog scene transmittance, algorithm to recover images clear natural color. But the original algorithm with high computational complexity, time consuming, can not meet the demand of real-time. Methods: This paper presents an improved algorithm for dark channel prior based, by improving the dark channel image calculation method, and the use of image guided filtering to optimize transmission diagram, phase comparison method. In this paper the previous method has been greatly improved in the algorithm processing speed, and get the good visual effect, at the same time, gives the analysis of the experimental results of 4 methods and evaluationResults: the dark color oriented filtering method based on image color distortion can be seen after treatment from the subjective evaluation, contrast balance, details accurately, from the objective evaluation is better than the other three methods, and the processing speed can meet the requirement of real time conclusion: compared with various algorithms, detail enhancement and color fidelity can achieve better results

結果 (英文) 2:

Abstract: objective: to single image to fog algorithm based on dark color a priori better solved the problem of estimating fog scenes transmission, algorithms for recovery image clarity color natural. But the original algorithm complexity, lengthy, failure to meet high real-time requirements. Method: this article made has a based on dark channel transcendental of improved algorithm method, through improved dark channel image calculation method, and used image Guide filtering to optimization transmission figure, makes this article made of method phase compared predecessors of method in algorithm handling speed Shang has has is big of increase, and get of good of image Visual effect, while to out has 4 kind of method of experiment results and evaluation analysis. The result: dark primary colors from the subjective evaluation based on oriented filtering can be seen undistorted image after color, contrast balance, detail restore accurate, objective evaluation methods are superior to the other three, while processing speeds to meet real-time requirements conclusion: compared with the various algorithms, detail enhancement, and color fidelity can be achieved better results.