Research On Electronic Image Stabilization Technology For Industrial Video Surveillance Based On Gyroscope

Machine vision technology is increasingly widely used in the industrial field,and camera equipment in the process of collecting video,due to factors such as vibration or external noise generated during the operation of complex heavy equipment platforms,causes jitter or distortion in the collected video images,seriously affecting the imaging quality and subsequent image processing and analysis,resulting in unstable visual detection and measurement,large errors,and other situations.Therefore,Image stabilization technology has broad application prospects in the field of industrial video surveillance.(1)Research on data preprocessing problems.This article establishes a reprojection error equation to calibrate the parameters of the sensor and calculates the delay time to solve the problem of time asynchrony;The collected gyroscope data was filtered and interpolated to ensure complete correspondence between video frames and gyroscope data,laying the foundation for subsequent preliminary and final image stabilization modules.(2)Aiming at the problem of video rotation jitter caused by industrial equipment platform vibration.This article uses a gyroscope sensor for estimation and compensation of rotation,overcoming the difficulty of traditional image stabilization algorithms greatly affected by image scenes.The relationship between the three coordinate systems is established through the camera calibration principle,and the estimation and reverse compensation of the rotation vector of the jittered video are completed.The experiment proves that the algorithm has a preliminary image stabilization effect on the jittered video.(3)Aiming at the problem of video translation jitter caused by industrial equipment platform vibration.This article applies the quadtree idea to the ORB algorithm to extract feature points,and then combines the BEBLID algorithm and PROSAC algorithm to more efficiently describe feature points and eliminate mismatched point pairs.Through affine motion model parameters,the translation jitter vector of the camera platform is accurately estimated and compensated.Finally,image stitching technology is used to solve the problem of edge pixel loss in the reverse compensation process.On the basis of preliminary image stabilization,final image stabilization is carried out,Output stable,smooth,and clear video images.Compared with traditional image stabilization algorithms,the algorithm proposed in this paper improves the speed and accuracy of the image stabilization system,and does not rely on image scene information.The experimental results show that the image stabilization algorithm used in this paper has improved the PSNR value,SSIM value,and execution efficiency of the image stabilization system compared to traditional image stabilization algorithms,Effectively solves the problem of unintentional shaking and distortion of video images collected by the camera system due to industrial equipment platform vibration or external noise interference,providing clear and stable high-quality video images for subsequent image processing in the production and operation of complex heavy equipment.