Research Of Digital Video Stabilization Algorithm For Tower Crane Hook Monitoring

The tower crane hook visualisation monitoring system transmits real-time hook video images to the tower crane driver by installing camera devices on the crane boom trolley,effectively solving the problem of blurred long-distance vision and visual dead space for tower crane drivers under complex construction sites such as blind cranes and isolated mountain cranes.However,in the actual application process,due to the camera device by the tower crane movement and external environmental factors,resulting in the collection of hook video image shaking,unclear,shaking video will not only cause visual fatigue to the tower crane driver,and will lead to driver misjudgment,missed judgment.This paper combines video characteristics of tower crane hook and uses digital video stabilization algorithm to stabilise the hook video,which has important engineering application value.The main work of this paper is as follows:(1)For the problems of high resolution and uneven light distribution of tower crane hook video image,the image is pre-processed with downsampling and grey scale histogram equalization algorithm to reduce the image resolution and improve the uniformity of image grey scale distribution,which provides a good data basis for the subsequent motion estimation and reduces the computation amount of data processing at the same time.(2)For the problems of inaccurate and inefficient motion estimation caused by the complex background of tower crane hook video and interference from foreground targets,this paper proposes a motion estimation method combining the optimized ORB algorithm and background compensation algorithm.As the classical ORB algorithm detects feature points which are prone to problems such as aggregation and redundancy,an image chunking and adaptive thresholding feature point detection method is used to optimize the classical ORB algorithm,and an image quadratic tree algorithm is introduced to improve the uniformity of feature point distribution,based on which the BEBLID algorithm is used for feature point description,and the k-nearest neighbour algorithm and PROSAC algorithm are used for coarse and fine matching of feature points.Considering the influence of foreground motion targets in the hook video on the accuracy of motion estimation,the background compensation algorithm combined with the inter-frame difference method is used to quickly identify foreground targets and reject them,which improves the accuracy of global motion parameter estimation.(3)For the problem of random noise in global motion parameters,this paper adopts fixedlag Kalman filtering algorithm to remove the random jitter components in global motion parameters.The algorithm not only meets the real-time requirements,but also effectively removes the high-frequency random jitter in the original motion traj ectory,and accurately obtains the image compensation parameters,thus generating a stable and clear video of tower crane hooks.(4)The video stabilization algorithm of the tower crane hook in this paper was applied to the construction site for experimental analysis in different operating scenarios.The experimental results showed that compared to this video stabilization algorithm that combines classical ORB and Kalman filter,the algorithm in this paper has better vidoe stabilization effect;The ITF value of the video stabilized by the algorithm in this paper has increased by an average of 50.37%compared to the original video,and the SSIM value has increased by an average of 19.28%,achieving good image stabilization effect,and the FPS value has reached over 29 frames/s,meeting the real-time monitoring requirements of the tower crane hook.