Image Invariant Feature Extraction Method Based on Circular Transformation

Abstract

The invention relates to an image invariant feature extraction method based on circular trace transformation, which belongs to the field of image processing and computer vision. This method uses a new circular trace transformation to track the image, and maps the functional result on the circular trace to the space generated by the three parameters of radius, length and angle to obtain the circular trace transformation result. Continuing to do functional integration on this result, the image quadruple circular track spatial features can be obtained. Using different functionals on the circular trace line can obtain different circular trace transformation results, and different functional combinations can obtain different circular trace transformation features, that is, it can represent different texture properties of images, and generate multi-dimensional circular trace transformation texture features information. This method uses the texture features extracted by circular trace transformation, which has RST invariant characteristics, can obtain deeper image information, and has better resolution ability for images containing circular and arc-shaped textures. In the case of fewer training samples The recognition ability has been significantly improved.

Description

technical field [0001] The invention relates to an image invariant feature extraction method based on circular trace transformation, which belongs to the field of image processing and computer vision. Background technique [0002] Image feature extraction plays a key role in image processing. How to extract features that reflect the essence of the image and have strong adaptability from the image has always been the core research content in the fields of image processing and computer vision. As we all know, the most common image features include texture, color, and shape. In the process of extracting these features, researchers try to find some invariant features. When , these feature quantities are invariant. Image invariant feature analysis methods mainly include moment theory analysis method, correlation analysis, Fourier descriptor, autoregressive model, SIFT (Scale invariant feature transform, scale invariant feature transform), etc. The generalized analysis method, t...

AI Technical Summary

Problems solved by technology

[0004] Although many researchers have done a lot of research on the invariant feature extraction based on trace transformation, most of them are on the combination optimization of trace transformation functionals, analysis and research on the rotation invariance and scaling sensitivity of different functionals, and trace transformation. In terms of transformation application research, there are few studies on invariant feature extraction methods for images with different geometric textures, such as images with circular or arc-shaped textures.

Images