Image stitching is one of the successful applications of computer vision. This technology is applied and optimized in vehicle undercarriage scanner, a sub-system of Eliteun MVSS (Multi-factor Vehicle Surveillance System). The optimized functions enable Eliteun system to scan undercarriage of vehicle in motion at a speed up to 30km/h.
Image stitching is the process of combining multiple photographic images with overlapping fields of view to produce a segmented panorama or high-resolution image. The process briefly refers to image preprocess, image alignment and image fusion.
Images of the same object but captured from different angles would differ in position, brightness or in contrast,
(Pic.1) affecting extraction of feature points, crop of reference images, and composition of the final image. In the scenario of undercarriage scanning, the height of which would influence significantly on brightness and contrast of snapped photos (Pic.2), thus posing great challenges on algorithm development, optimization and engineering.
As the core of the entire process, image alignment determines how quickly and accurately the final image is stitched. Specific algorithms are applied to find and match feature points in two images. (Pic.3) Parameters can be calculated either by pixel-to-pixel contrast or by gradient descent algorithm, to finally complete image alignment.
In Eliteun MVSS undercarriage scanner, sub-images are aligned, cropped along the recognized boundary, then sent to back-end server to compose a complete and clear image of a vehicle undercarriage. (Pic.4)
Sample image 1&3 source: https://www.cnblogs.com/skyfsm/p/7411961.html