Research on Image Stitching for Parking Assistance System

Currently, car parking assistance is limited to the small angle rear view imaging of the reversing radar, which can only provide the driver with a limited range of vision and is prone to safety hazards. It is necessary to stitch the surrounding images captured by the parking monitor by widening the monitoring field of view of the car parking and providing the driver with a complete rear panoramic image through the parking assistance system. In this research, we provide an enhanced SIFT image stitching technique that uses random k-d trees and the k-nearest neighbour matching to optimise SIFT feature matching to improve matching speed, the RANSAC algorithm to reject mismatched pairs to improve matching accuracy, and pre-processing to reduce distortion and eliminate stitching gaps by applying columnar projection and weighted fusion to the stitched images. The experimental findings demonstrate that the suggested approach may increase image stitching speed while also producing stitched and displayed images that are more accurate.