Image Fusion of Catenary Components Based on Laplacian Pyramid

Aiming at the problems such as the lack of information in traditional images, the loss of color and details in polarization images, and the loss of some surface details, a polarization image fusion technique based on the Laplacian pyramid of catenary components is proposed. Firstly, the acquired polarization image is aligned and denoised, and then the polarization image and light intensity image are obtained by Stokes' formula, and then the Laplacian pyramid transform is used to decompose it. In the third level, low-frequency component is fused by the fusion strategy combining the variance matching degree and the regional variance adaption. For the low-frequency components in the first and second levels, a weighted averaging strategy is employed for fusion; and the fusion of the high-frequency components are fused by the strategy of taking the pixel to the larger one. Finally, the fused image is subjected to adaptive histogram equalization to enhance the image. The results of the trial experiments show that the details of the image obtained by fusion of this method are more obvious, the fusion effect is better, and compared with other traditional image fusion algorithms, the objective evaluation indexes of this method are all higher than those of other methods, which provides a reference for the detection of the railroad system.