Hyperspectral target tracking method based on joint spectral dimension reduction and feature fusion

Abstract

The invention provides a hyperspectral target tracking method based on joint spectral dimension reduction and feature fusion. The hyperspectral target tracking method comprises the following steps: firstly, performing dimension reduction processing on an original hyperspectral image sequence by utilizing a joint spectral dimension reduction method based on multi-dimensional scaling and principal component analysis; respectively extracting four pairs of features of the image sequence obtained after dimension reduction processing, and fusing the four pairs of features; sending the fused featuresto a kernel correlation filter, and obtaining four weak response graphs based on the first to fourth features; weighting the weak response graph by using the weight coefficient to obtain a strong response graph; taking the position of the maximum value in the strong response graph as the position of the target; and adaptively updating the parameters of the base sample and the weight coefficient.According to the invention, the defects of large calculation amount and poor real-time performance in the prior art are overcome, the target tracking speed in the hyperspectral image sequence under the complex background is increased, and a good tracking effect is achieved when the target deforms and is shielded.

Description

technical field [0001] The invention belongs to the technical field of image processing, and in particular relates to a hyperspectral target tracking method based on joint spectral dimensionality reduction and feature fusion. Background technique [0002] Target tracking in hyperspectral image sequences under complex background is an important part of hyperspectral image processing technology. practical application. In recent years, target tracking methods based on improved kernel correlation filtering have been widely used in the field of computer vision. The kernel correlation filtering algorithm uses the gray features of the basic samples to track, but the gray features of the hyperspectral target are not enough to distinguish the complex background from the target in the background. [0003] In the existing target tracking method, by extracting the spectral features in the target search area, training the spectral correlation filter Spe-CF, selecting the target search ...

AI Technical Summary

Problems solved by technology

The disadvantage of this method is that this method processes hyperspectral video of all bands simultaneously, and uses the spectral correlation filter Spe-CF and spatial correlation filter Spa-CF to estimate and determine the target position, and pre-training is required Filter, with a large amount of calculation, poor real-time performance, and when the target is occluded and deformed, it is easy to fail to track

The shortcomings of this method are: it is necessary to build a target sample library to train the deep convolutional network, which requires a large amount of calculation, and the algorithm is easily affected by target occlusion and deformation, resulting in estimation errors and tracking deviations.

Images