Depth Map Information from Stereo Image Pairs using Deep Learning and Bilateral Filter for Machine Vision Application

Stereo matching algorithm is a subset of machine vision study, and it is effective for creating accurate depth maps that are utilized in a variety of applications. It is found that the most difficult challenge for the stereo matching algorithm is obtaining an exact corresponding point in low texture regions. This study offers an approach that uses deep learning (DL) and bilateral filter (BF) to improve depth map precision in this area. The application of an edge-preserving filter such as this is capable of refining and removing excessive noise on the resulting image, as the filter is resistant to high contrast and brightness. Based on experimental research utilizing a common benchmarking dataset from Middlebury, the suggested method in this study provides good accuracy and is competitive with other published methods.