Human body tracking method based on self-adaptive kernel function and mean value shifting

Abstract

The invention relates to a human body tracking method based on a self-adaptive kernel function and mean value shifting. The human body tracking method includes two stages, the first stage is a learning stage, a set of training samples of human body walking is firstly read, human body prospect shapes are mapped to be coordinates in a low-dimensional space through a dimensionality reduction algorithm, a low-dimensional human body shape space is obtained, the human body prospect shapes are then recovered through an interpolation reconstruction algorithm, and parameters, capable of mapping from a low dimension to a high dimension, of the interpolation reconstruction algorithm can be obtained. The second stage is a tracking stage, a human body optimum kernel shape in a video frame is searched for in the low-dimensional human body shape space, and the human body in the video frame is tracked by using a mean value shifting algorithm. Compared with the prior art, the human body tracking method improves the shape of the kernel function in a traditional mean value shifting algorithm, so that the shape of the kernel function is not fixed and changes in a self-adaptive mode according to changes of shapes of the tracked human body, histogram modeling and matching of the kernel function are further performed in the high dimension space, and therefore the performance of a human body tracking technology is improved.

Description

technical field [0001] The invention relates to the technical field of image processing and computer vision, in particular to a human body tracking method based on an adaptive kernel function and mean value shift. Background technique [0002] Object tracking is an important branch in video supervision, which has achieved great progress in recent years. At present, in the process of target tracking becoming practical, many problems have been encountered, such as object occlusion, low background contrast, complex object movement, etc. The most important problem is the human body tracking problem, because the human body is a non-rigid object structure. , that is, objects with deformation, especially when the human body is moving, the shape of the human body changes, which greatly increases the difficulty of target tracking. difficulty. [0003] In recent decades, many researchers at home and abroad have begun to study target tracking. These methods can be divided into three ...

AI Technical Summary

Problems solved by technology

[0006] The self-deformation of the target means that when the target is a non-rigid object and usually has its own deformation when it moves, such as a human body walking, at this time, the symmetric regular kernels in the traditional mean shift algorithm, such as rectangular kernels, elliptical kernels, etc., can no longer accurately describe the object When tracking a human target, the elliptical or rectangular kernel function of the traditional mean shift algorithm contains points in the background area, and these points will act as the foreground points of the human body in the tracking process, which will lead to inaccurate tracking and positioning , there are problems such as mistracking and lost tracking targets

In addition, the traditional mean shift algorithm can only track and locate the human target, but cannot accurately describe the shape of the human body

Images