A method for classifying gabor features of hyperspectral images based on incremental local residual least squares

Abstract

The invention discloses a hyperspectral image Gabor feature classification method based on incremental local residual least squares, comprising the following steps: reading hyperspectral image data cube h(x, y, b); using given parameters (frequency Amplitude and space spectrum space direction) Gabor filter filters the hyperspectral image cube to obtain the feature image cube; obtain the pixels of the marked category in the feature image cube as the training pixels, and the pixels of the unmarked category are the test pixels; Under the current Gabor feature, use the incremental local residual least squares method to update the reconstruction coefficient matrix sequentially, and the cumulative class local residual between the reconstructed value of the test pixel and the real test pixel; clear the current Gabor feature image data cube; in the next step Under the given Gabor filter parameters, a new Gabor feature data cube is extracted, and the above update steps are repeated to obtain the final category local residual after traversing the given Gabor feature parameter set; classification is performed according to the minimum category local residual criterion.

Description

technical field [0001] The invention relates to the field of image processing, in particular to a hyperspectral image Gabor feature classification method based on incremental local residual least squares method. Background technique [0002] Hyperspectral images have the characteristic of containing both spatial information and spectral information of the observed scene, and the coexistence of spatial and spectral information makes hyperspectral images have a wide range of practical application prospects and theoretical research value in many fields. Gabor filtering can effectively extract local frequency domain information in space-spectrum space from hyperspectral images. In practical applications, it is often necessary to connect and combine Gabor features under multiple sets of parameters to obtain new high-dimensional features, resulting in a large scale of data. Calculation poses certain difficulties. [0003] There are many hyperspectral image classification methods ...

AI Technical Summary

Problems solved by technology

However, when extracting high-dimensional Gabor features from hyperspectral images, the classic least squares method is prone to insufficient memory due to the large amount of data to be processed, which affects the actual reliability of Gabor feature least squares classification. operability

Images