Tattoo image classification method based on deep learning

Abstract

The invention discloses a tattoo image classification method based on deep learning. The tattoo image classification method based on the deep learning comprises the following steps: 1) transforming samples: 1.1) affine transformation, 1.2) elastic transformation, 1.3) sheltering simulation and 1.4) whitening; 2) carrying out self-encoding pre-training: training a great quantity of colorful tattoo images by a self-encoding learning machine which is optimized by CUDA (compute unified device architecture) to obtain certain common edge information of the tattoo images, and meanwhile, picking up and applying the images to the first layer of a convolutional network; 3) carrying out convolutional network training to a transformed sample by a result obtained by self encoding. The image classification method based on the deep learning, which is disclosed by the invention, is effectively free from the influence by irradiation direction, skin color, hair, ray, image quality and the like, and has the advantages of good reliability and high realization efficiency.

Description

Technical field [0001] The invention relates to the technical fields of image processing and pattern recognition, in particular to a tattoo image classification method. Background technique [0002] The research on tattoo image recognition has just started, and there are no domestic related papers and patents. In foreign countries, only a few scholars such as A.K. Jain are engaged in tattoo-related academic research. Although there have been certain developments, there are still many problems that need to be solved urgently. First, the existing methods for detecting regions of interest have certain defects. The detection of the interest area of ​​existing tattoo images is mainly through manual calibration or image segmentation algorithms. The image segmentation algorithm is an estimation of the actual interest of users by low-level visual features, and it is difficult to achieve an objective description of the area of ​​interest. Therefore, both methods have subjective issues....

AI Technical Summary

Problems solved by technology

[0022] In order to overcome the shortcomings of the existing tattoo image classification methods, which are easily affected by the direction of irradiation, skin color, hair, light, image quality, etc., have low reliability, and low implementation efficiency, the present invention proposes a method based on deep learning. Tattoo image classification method, which relies on a super large sample set, pre-processes the image of the sample, uses an autoencoder to pre-train the color tattoo image, and then uses a convolutional network for learning; considering the low efficiency of the convolutional network in CPU operation , the present invention uses CUDA-optimized convolution, which improves the efficiency by 40 times compared to the CPU, and the learned results are used in the actual application of tattoo classification, with good reliability

Images