Semi-Global Stereo Matching Algorithm Based on Optimized Image Preprocessing

This paper proposes an improved Semi-Global Matching algorithm that enhances the quality of disparity images by optimizing image preprocessing.The method applies denoising, contrast enhancement, and Sobel filtering to compute pixel gradients and highlight image edges, thereby improving image quality and reducing noise interference to enhance the clarity of feature boundary.The preprocessed images are used for the SGM algorithm, which improves matching accuracy and adaptability through adaptive sliding windows and dynamic aggregation strategies for cost calculation and aggregation.Experimental results indicate that the improved algorithm enhances the accuracy and robustness of disparity images.