Object Position Tracking Method Based on Correlation Filter Algorithm
A filter and correlation technology, applied in instrumentation, calculation, image analysis, etc., can solve the problems that the correlation filter algorithm cannot judge the tracking results, cannot produce ideal object tracking results, and wrong tracking results, etc., to achieve Avoidance of degradation problems, high reliability, effect of reducing influence
Active Publication Date: 2021-02-12
ZHEJIANG UNIV
View PDF7 Cites 0 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
However, the correlation filter algorithm cannot judge the quality of the tracking results, which will cause serious error accumulation problems
Secondly, when the tracked object is occluded or severely deformed, the correlation filter algorithm will directly produce wrong tracking results and directly affect the subsequent object tracking
In addition, the correlation filter algorithm uses a single learning rate value to update the underlying model, which has poor adaptability to target objects and changes in the surrounding environment, and cannot produce ideal object tracking results in complex scenes
Method used
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View moreImage
Smart Image Click on the blue labels to locate them in the text.
Smart ImageViewing Examples
Examples
Experimental program
Comparison scheme
Effect test
Embodiment Construction
[0017] The present invention will be further described below in conjunction with the accompanying drawings and specific embodiments.
[0018] Such as figure 2 As shown, the present invention performs object position tracking based on the correlation filter algorithm, and the specific steps are as follows:
[0019] 1. If figure 1 As shown, M is the currently input image frame. The standard Kalman filter is used to construct a kinematics predictor to predict the two-dimensional position p of the target object in the current image, which is called the estimated position. The state variables of the Kalman filter include the two-dimensional position coordinates of the target object and the movement speed in the direction of the horizontal and vertical axes of the image, and the measurement variables of the Kalman filter are the two-dimensional position coordinates of the target object. In the process of pre-estimation calculation, the motion of the target object is considered t...
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More PUM
Login to View More Abstract
The invention discloses an object position tracking method based on a correlation filter algorithm. The present invention uses Kalman filter to construct a kinematics predictor to predict the position of the target object. Combined with the previous movement of the target object to estimate the current position, and then combined with the measurement results generated by the correlation filter algorithm to correct the estimated position, so as to produce a smoother trajectory and greatly reduce noise interference . The present invention adopts the response quality evaluation model to measure the reliability of the object tracking result of the correlation filter algorithm, and directly adopts the estimation result of the kinematics predictor under unreliable conditions, and does not update the underlying correlation filter, avoiding correlation Degenerate problem of linear filter algorithm. The invention has good accuracy, robustness and high reliability, and can effectively track the position of any object in a two-dimensional image.
Description
technical field [0001] The invention relates to an object tracking technology in machine vision, in particular to a real-time object tracking technology based on a correlation filter algorithm. Background technique [0002] Object tracking is a very challenging research problem in the field of machine vision. Given the position information of the target object in the initial image frame, its position information in each subsequent image frame is solved through a certain mathematical model and algorithm analysis. Object tracking has great application value in security monitoring, unmanned aerial vehicle tracking, automatic driving, traffic flow control, intelligent robotic systems, augmented reality, and human-computer interaction. The correlation filter algorithm is an online training classifier model in the frequency domain. It has been successfully applied to the object tracking problem in recent years. Its underlying mathematical principle is simple, it runs fast, and it...
Claims
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More Application Information
Patent Timeline
Login to View More Patent Type & Authority Patents(China)
IPC IPC(8): G06T7/246G06T7/277G06T7/73
CPCG06T2207/10016G06T2207/20081G06T2207/20084G06T7/246G06T7/277G06T7/73
Inventor 任重张家玄
Owner ZHEJIANG UNIV