基于CGAN的EMT钢轨伤损检测图像处理方法

Image Optimization Algorithm Based on CGAN for Reconstructed Images of Rail Damage by EMT

摘要: 为了提高基于电磁层析成像技术（electromagnetic tomography，EMT）的钢轨伤损检测重建图像的质量，提出了基于CGAN（conditional generative adversarial networks）的EMT钢轨伤损检测图像处理方法. 通过在电磁仿真软件中获取仿真数据，使用传统EMT成像算法LBP（linear back projection）对钢轨伤损进行初步成像；使用CGAN对初步成像的钢轨伤损图像进行处理；对比处理前后图像的成像效果，发现处理后的钢轨伤损图像更为清晰和准确，图像评价指标有较大提升，更加适用于钢轨伤损检测；另外对算法的泛化能力也进行了分析.

Abstract: In order to improve the quality of the reconstructed images of rail damage by electromagnetic tomography, image artifact processing algorithm based on conditional generative adversarial networks (CGAN) was proposed. Simulation data was obtained by electromagnetic simulation software, and the images reconstructed by traditional EMT reconstruction algorithm LBP was used as input of CGAN for image processing. The image quality before and after processing and the corresponding objective image evaluation index were compared to verify the effectiveness of the proposed image processing algorithm based on CGAN, and the generalization ability of the proposed algorithm was analyzed too.