A target scale adaptive tracking method based on OpenCV

Abstract

The invention discloses a target scale adaptive tracking method based on OpenCV. The method comprises the following steps of: positioning and tracking a target by utilizing a mean shift algorithm; extracting target features by using SIFT (Scale Invariant Feature Transform) every 6 frames; carrying out feature matching on the template image and the tracking real-time image, and determining a mapping relation between matching points through affine transformation to obtain a scale transformation factor, thereby updating the size of a candidate region target template in the real-time image, and solving the problem that tracking is easy to lose or track mistakenly because the scale of a moving target is continuously increased.

Description

technical field [0001] The invention relates to the technical field of computer vision, and specifically refers to an OpenCV-based target scale adaptive tracking method. Background technique [0002] Moving target tracking is an important research direction in the field of computer vision. It has broad application prospects in video surveillance, automatic navigation, space remote sensing, infrared medical image pathological analysis, urban infrared pollution analysis, forest fire prevention, and sea surface personnel search and rescue. [0003] Target tracking methods mainly include two categories: 1) tracking methods based on state estimation; 2) tracking methods based on matching. The state estimation-based methods mainly include Kalman filter and particle filter, and the matching-based tracking methods mainly include mean shift, SIFT (Scale Invariant Feature Transform), etc. The mean shift tracking method is a non-parametric fast pattern matching method based on kernel ...

AI Technical Summary

Problems solved by technology

However, the SIFT algorithm is very dependent on the content and quality of the image, and the infrared image has blurred visual effects, low resolution, and much noise, and it is difficult to extract enough feature points

As a result, the tracking algorithm cannot continue, and the implementation of the algorithm is complex and the amount of calculation is large, which hinders the real-time tracking of the target

[0005] So far, many scholars have improved the mean shift algorithm, such as using plus or minus 10% increments to correct the kernel window width, but when the target gradually increases, especially when the target size exceeds the kernel window width, the kernel The window width is difficult to expand, but often becomes smaller

The problem existing in the existing technology is that it is not accurate enough to rely on only a single or small amount of information as the tracking standard, and it is easier to lose track or mistrack when the target scale continues to increase

Images