改进的偏微分扩散方程图像增强方法

通过改进P-M扩散系数和融入相干增强扩散改善了传统的P-M模型，使得裂缝图像非边缘处以改进后的P-M扩散为主，在边缘处以相干增强扩散为主，达到了对裂缝图像去噪和增强的目的．理论分析和仿真实验表明，提出的方案较传统的P-M模型能使地面图像的特征得到更好的保护，使裂缝更加明显，利用该方法对裂缝图像进行预处理后，提高了路面裂缝目标检测的准确性，体现了该方法的优越性．

Image enhancement using the improved partial differential equation

Excellent preprocessing is necessary for the crack extraction from pavement images. This paper adopts the Partial Differential Equation method to do that, and then improves the P-M diffusion coefficient and fuses it with coherence enhancing diffusion, thus forming the new PDE model. The new PDE model makes the improved P-M diffusion dominates the region without crack edges and coherence enhancing diffusion close to the crack edges, thereby realizing the crack image denoising and enhancing in a more effective way. Theoretical analysis and simulation results consistently show that the proposed model outperforms the classical P-M model in terms of pavement image denoising and enhancing. Furthermore, after the pavement images are preprocessed by the proposed PDE model, cracks are detected more accurately, demonstrating the superiority of the proposed preprocessing method.