A Research About Algorithms Of Automatic Jacquard Image Stitching

With the process of digitalization, digital image processing has made a great progress in all aspect of our lives. Among the technologies of digital image processing, image stitching is widely applied to the pattern recognition, computer vision, virtual reality technology, medical, military and other areas as a hotspot.Targeting at the stitching needs of jacquard scanned images in jacquard CAD, the thesis compares and analysis the advantage and weakness of varies methods which are studied all around the world. Then designs and improves a method of image stitching which is suit for jacquard scanned image. Firstly, an improved SIFT(Scale Invariant Feature Transform) algorithm is used to locate and match the feature points in the scanned images. Then a special algorithm is employed for getting the best matching points to compute the transform matrix, and the overlapped area linear transition method is incorporated for seamless stitching. Finally, finds the least error of transform matrix in stitching of any multiple images to confirm the best standard space which is used for image mapping.The jacquard images automatic stitching system in the thesis is a replace of a similar artificial function in normal jacquard CAD. Meanwhile, this system can achieve more stable stitching effects and better image qualities. It can be applied to any images with various sizes, luminance, and rotations, which further improved the jacquard CAD.