Research Of Single Image Haze Removal Based On Physical Model

Fuzziness of atomization image causes very bad interference for the outdoor video surveillance system and people’s life.For this problem,a lot of methods were adopted to make atomization image clear.However,all this methods have insufficient.So new solution need to be putted forward.Benefiting from substantial digital image processing technology,in the theoretical framework of physical model of haze image and dark channel prior and with the goal of improving the image contrast and repair the pixel color,this article use the combinational method of minimum filtering and guide filtering to repair the transmittance and use a parameter to increase intensity of the dehazing image in image haze removal.In addition,this paper analyzes the failure of dark channel prior used in the white sky area.To solve this problem,we put forward an effective solution.The content of this article as follows:1、To determine the image processing theoretical framework for single image dehazing and explain the physical model.Image defogging is belonging to image enhancement with the perspective of just raising the contrast of atomization images.Considering the formation reasons of haze image and building physical model for it,we can use image restoration method to defog image.Image enhancement method can’t restore the original color value of image and a specific enhancement algorithm is not universal application.So this article adopts image restoration method for haze image removal and tries to make it clear.Physical model which describes the degraded mechanism of atomization image needs to be built.Then the mathematical relationship between atomization image and clear image will be found.So solving the mathematical equation of physical model is identical to image haze removal.2.Eliminate white halo in the depth edgesTraditional dark channel prior method produces white halo in the depth edges.He uses soft matting to optimize the transmittance.It gets good result,but it takes plenty of times and is of high complexity.In the paper,we fully analyze the advantages and disadvantages of minimum filtering and guide filtering respectively.Then,we adopt the combinational method of minimum filtering and guide filtering,which combine the advantages of them.The algorithm not only obtains good result of eliminating white halo but also reduce the complexity.At the same time,according to the experiment result,this article fully analyzes the dehazing results by different weights of minimum filter and guide filter and reach relevant conclusions.3、Increase the intensity of dehazing image and evaluate algorithm’s performance.Intensity of dehazing processed by dark channel piror is a little low.Intensity increasing parameter is used in the image haze removal,so that we can increase theoverall brightness of the restored image and improved visual effect.Except subjective judgment,this text use contrast and complexity to evaluate algorithm’s performance objectively and design a formula to calculate contrast.4、Process white sky area in the fog image.Dark channel prior fails in the white sky area of the fog image.It causes bad visual effect and color distortion.According to the bright white features of pixels in the sky area,a useful solution was putted forward to solve it based on the sky color transcendental plane and its normal vector.It effectively improves the visual effect.