Research And Implementation Of Video And Image Stitching System

Video monitoring technology is widely popular in daily life since this century.Serving as an assistant to the police,it can monitor the crime in real time and provide evidence of the first scene afterwards.However,due to cameras with a limited range,in order not to reduce the resolution,we shoot wide videos with image stitchings.In fact,video stitchings consist of image stitchings in every frame.There are many drawbacks in sitching method.On the one hand,traditional methods represented by SIFT and SURF perform not well on the time complexity and the image edge details.On the other hand,the new algorithm represented by the ORB is less robust in scale invariance and uneven distribution of feature.Therefore,we put forward an image registration algorithm called NLORB and a blend method with brightness,based on the nonlinear scale space and the ORB,and then we carry on the verification in this paper.We describe the main tasks as follows:(1)In order to solve problems of distribution of scale invariance,feature and time complexity in traditional methods,we put forward an image registration algorithm based on the nonlinear scale space.(2)Based on the projection transformation of the cylinder,we transform different image surfaces to the same coordinate system.We blend images with brightness template and present the result.(3)With FFmpeg library as the underlying storage service,we realize the function of pulling,decoding,encoding and pushing(4)We design interface and play videos in multi screens with Qt library,and eliminate noise through image preprocessing with OpenCV library.(5)We program to implement the video and image stitching management system and the whole system is optimized and accelerated.In order to evaluate the feasibility of this method,we build a video and image stitching system in this paper.The test results show that the system can realize the basic functions and provide necessary reference for future research.