Visual tracking method based on convolution feature and manual feature integration

Abstract

The invention discloses a visual tracking method based on convolution feature and manual feature integration. The method comprises the following steps: extracting the depth feature of a target image through VGG19, and employing a conv3-4 layer, a conv4-4 layer and a conv5-4 layer as the three-layer convolution features of the target image; obtaining HOG features, CN features and gray features, and fusing into a layer of manual features; obtaining the Fourier transform result of the three-layer convolution features and the Fourier transform result of the one-layer manual features, obtaining the target estimation position of the target image through calculation based on a correlation filtering algorithm, and completing visual tracking; and selecting a part of image frames in the current visual image through the interval parameter, taking the selected part of image frames as a new target image, and starting to update visual tracking. According to the invention, the accuracy and success rate of visual tracking are improved, and the running speed of the tracker is improved.

Description

technical field [0001] The invention relates to the field of computer graphics and target image processing, and relates to a visual tracking method based on convolution feature and manual feature integration. Background technique [0002] With the rapid development of computer vision technology, video-based object tracking algorithms have become a research hotspot in research institutions and universities at home and abroad. Target tracking technology usually builds a robust model based on the target and background information in the video to predict the shape, size, position, trajectory and other motion states of the target in the video, so that behavior prediction, behavior understanding and analysis can be performed. The current application fields of visual tracking are very extensive, including many fields such as video surveillance, unmanned aerial vehicle, military precision guidance, intelligent transportation and human-computer interaction, and have important researc...

AI Technical Summary

Problems solved by technology

[0004] Aiming at the above-mentioned deficiencies in the prior art, a visual tracking method based on convolutional features and manual feature integration provided by the present invention solves the problems of low visual tracking accuracy, low success rate and low speed

Images