A Target Tracking Method Based on Statistical Features of Orientation Gradient

Abstract

The present invention provides a target tracking method based on directional gradient statistical features, comprising the following steps: selecting an initial target template image and extracting directional gradient statistical features; calculating the frequency domain features of the initial target template image; acquiring search area images and extracting directional gradient statistics Features; calculate the frequency domain features of the image in the search area; obtain the update position, scaling and rotation angle of the tracking target through the frequency domain response; select the current target template image and extract the direction gradient statistical features; calculate the frequency domain features of the current target template image; use the current target The frequency domain feature of the template image updates the frequency domain feature of the initial target template image, and performs target tracking on the next frame image. The invention effectively weakens the interference of factors such as image brightness, contrast, target deformation, target rotation, etc., realizes precise tracking of the target, and simultaneously improves the operation processing speed, thereby improving the real-time performance of target tracking.

Description

Technical field [0001] The present invention relates to target tracking feature, and in particular relates to a method of tracking a target direction based on the statistical characteristics of the gradient. Background technique [0002] Characteristics of the target complex background has been the difficulty of tracking the image tracking. Typically, there is a contrast area associated tracking and tracking method based on image matching based tracking algorithm. [0003] Tracking method to track contrast, the use of contrast difference between the target and background to identify and extract the target signal, so as to achieve target automatic tracking method. This method of tracking according to the tracking reference points can be divided into: edge tracking, centroid tracking peak tracking, can quickly track the target, the target posture change adaptability, but the difference in the ability to identify the target, it is difficult to track complex background the target app...

AI Technical Summary

Problems solved by technology

Because this algorithm uses the global information of the target, such as color, texture, etc., it has high reliability, but at the same time it has obvious limitations, such as being susceptible to changes in the light and shade of the image background, the distance scale transformation of the target, and the rotation and other factors, and the calculation load of the algorithm is very large. Although there are many simplified and optimized search algorithms, in special cases (high-resolution images or targets with relatively large areas), the real-time performance of the algorithm hard to guarantee

Images