Real Time Continuous Image Stitching Algorithm Based on SIFT

Image stitching technology has application scenarios in many fields. At present, the existing algorithms usually use the method of acquisition first and then synthesis, which is still insufficient in real-time performance; in addition, under the premise of real-time acquisition, it is also insufficient to realize continuous image synthesis. To solve this problem, this algorithm uses the RGB module of the Intel Realsense D435 camera for image acquisition, first creates an original image for storing the final result, collects an image at an interval of 100ms each time, and performs 8*8 hashing of the two images before and after Compare the similarity of the value, take the image with a similarity less than 5/8 and keep it, then use the SIFT (Scale-invariant feature transform) scale-invariant feature transform feature detection algorithm to extract the image feature points from the collected image, and use the RANSAC to extract the feature points The (Random Sample Consensus) algorithm screens effective points and calculates the homography transformation matrix at the same time. Finally, every two images collected are synthesized into one and covered in the corresponding position of the result image. Through experiments in this paper, the average time for single acquisition and synthesis is 70ms, achieving the real-time goal. The similarity between the experimental group and the control group can reach 70%, and the resolution is increased by 1.65 times, achieving the goal of continuous splicing.