Ear image recognition method amalgamating wavelet analysis and matrix feature

Abstract

The invention relates to an ear image recognition method combining wavelet analysis and moment features, which is characterized in that the method comprises following steps: (1) preprocessing collected ear images: using wavelet transform modulus maxima to denoise for the collected ear images, and extracting ear image edge to gain wavelet decomposition modulus maxima edge images in various scales; (2) extracting characteristic value of the ear images: making use of an improved wavelet moment invariants arithmetic to calculate the wavelet moment invariants used for extracting the characteristic value of the ear images; (3) recognizing the ear: processing weighting and classifying the characteristic value of the ear images gained in the step (2), and recognizing the ear. The invention processes to extract the characteristic value of the ear via wavelet modulus maxima and the improved wavelet moment invariants arithmetic, and not only has a high calculation accuracy and a fast calculation speed, but also can improve discrimination of the collected ear images in the environment of poor illumination, illumination change and noise jamming.

Description

technical field [0001] The invention relates to a personal identification technology of human biological characteristics, in particular to a human ear image recognition method which combines wavelet analysis and moment features. Background technique [0002] In recent years, biometrics has attracted more and more researchers' attention. It plays an important role in almost everything from authentication to gate entry security. However, most of the biometric technologies at this stage have strict requirements on their working environment, thus limiting their scope of application. Therefore, researchers are working hard to find new biometric technologies. Human ear recognition is one of the new recognition technologies, and there are very few studies on it at home and abroad. Human ear recognition technology has considerable theoretical research value and practical application prospect due to its unique physiological characteristics and observation angles. It involves many ...

AI Technical Summary

Problems solved by technology

[0010] The above calculation formula needs to use coordinate transformation when calculating the wavelet moment invariant to transform the digital image from the rectangular coordinate system to the polar coordinate system, which will increase the conversion error after the transformation, resulting in inaccurate value of the wavelet moment invariant

Another method uses explicit wavelet to calculate the wavelet moment invariant, but this method is difficult to integrate mallat fast algorithm into it, and the calculation speed is low

Moreover, the recognition rate will also be affected by uneven illumination, changes in illumination intensity, and noise interference.

Images