An Image Feature Extraction Method with Illumination Robustness

Abstract

The invention discloses an image characteristic extraction method with illumination robustness. The method comprises the following steps of: 1, collecting and uploading image signals, and adjusting resolution; 2, extracting image characteristics: carrying out two-dimensional wavelet transformation, transforming a low-frequency component after the two-dimensional wavelet transformation in a logarithm space, carrying out convolution calculation by using k Gaussian filter coefficients and low-frequency wavelet coefficients in the logarithm space, carrying out weighted average on a deviation between a convolution result and a low-frequency wavelet coefficient in the logarithm space, carrying out grey value linear stretching on the weighted average result, carrying out the two-dimensional wavelet inverse transformation of three high-frequency coefficients in a wavelet domain and the grey values to be displayed, repeating the steps, collecting and processing L image signals, and extracting the characteristics of the processed image signals; and 3, synchronously outputting processing results. The method disclosed by the invention has the advantages of rational design, convenience for realizing, low investment cost, simpleness and convenience for operation, good image processing effect and strong practicability. The illumination transformation robustness is increased.

Description

technical field [0001] The invention belongs to the technical field of image processing, and in particular relates to an image feature extraction method with illumination robustness. Background technique [0002] At present, in the image feature extraction of illumination changes, there are usually the following methods to deal with illumination changes: [0003] 1. Based on traditional image preprocessing methods, such as histogram equalization, Gamma transformation, etc. This type of algorithm has low complexity and fast operation speed, but it is often difficult to achieve the desired effect in terms of performance; [0004] 2. The method of extracting feature points with illumination robustness, embossing the image and adopting the edge map representation method, etc. These methods can obtain better recognition results when the lighting conditions do not change much. However, when the lighting conditions change greatly, the effect is not very obvious; [0005] 3. Bas...

AI Technical Summary

Problems solved by technology

[0007] In summary, the image feature extraction methods in the prior art have defects and deficiencies such as poor robustness to illumination, poor classification and recognition effects, and limitations. When face recognition is required, under the condition of changing illumination, the image feature extraction method in the prior art leads to poor face recognition effect, which cannot well meet the needs of practical applications

Images