The Algorithms Of Fog-Degraded Traffic Images Enhancement Based On Fractional Differential

To improve the drivers' visual effect, which can ensure the drivers getting the road information under the condition of low visibility ,it is the guarantee of safe driving. The key technology to ensure safe driving in the haze environment is to improve and enhance the visual effect of vehicle images. In recent years, many scholars found that fractional differential can improve the high-frequency information, enhance the intermediate-frequency information and nonlinearly preserve the low-frequency information of signal. Therefore, fractional differential used in the area of image can not only preserve the weak texture detail information, but also enhance the edge and contour information, so as to achieve the purpose of enhancing the image quality. In this paper, the algorithm of fog image enhancement based on fractional differential is studied .The main work is as follows:(1) The backward interpolation approximation scheme of first-order differential based on Lagrange interpolation method is derived in this paper .According to the idea of Lubich, the approximation formula of first-order differential can be extended to fractional differential. The second order numerical approximation scheme of Riemann-Liouville fractional differential is obtained. And a numerical example is given to testify the accuracy of the scheme.(2) The construction principle of image enhancement operator based on fractional differential is introduced, according to the theory of two order numerical approximation scheme of Riemann-Liouville fractional differential put forward in the second chapter, the two order backward difference (TBD) operator is constructed for dealing with the traffic images of bad weather conditions. Considering the specific center of TBD operator, in order to achieve better texture enhancement effect,the developed two order backward difference (DTBD) operator is proposed by improving TBD operator. Compared with other kinds of operators, the DTBD operator has superiority both in visual effect and objective evaluation index.(3) It is difficult to manually give the optimal differential order for excellent image enhancement effect in terms of images captured in complex condition, and it is hard to ensure the enhancement of the target object while preserving the information of background pixels if the entire image is filtered by a fixed differential order. To address this problem, we segment the image into the object area and the background area according to the theory of Iterative threshold and the Otsu thresholding algorithm based on Markov Random Field, And a piecewise function combining different features of the pixels in the object and the background is constructed to determine the appropriate order of differential in the two areas adaptively. Thus, two new adaptive fractional order differential algorithm of image enhancement based on segmentation were proposed for dealing with traffic images in bad weather conditions. Experimental results show the effectiveness of the proposed algorithm.