An Efficient Content-Based Human Skin Image Classification Retrieval Method

Abstract

The invention relates to an effective content-based human skin image classification and retrieval method, and belongs to the technical field of content-based image classification and retrieval. The method includes the following steps: S1: using a capacitance-based Fingerprint Sensor to collect human skin pictures; S2: using the 2D Gabor wavelet algorithm to classify and retrieve the collected human skin pictures, and analyzing the pictures with multiple channels and multiple resolutions, Extract the feature vector of the picture, and then compare the similarity of the feature vector between the pictures in the database, and finally complete the classification retrieval of the picture. The method provided by the invention provides good technical support for the technical field of content-based image classification retrieval, especially the field of human skin classification retrieval, and has broad application prospects.

Description

technical field [0001] The invention belongs to the technical field of content-based image classification and retrieval, and relates to an effective content-based human skin image classification and retrieval method. Background technique [0002] Human society in the 21st century has entered a digital age of information explosion. Through the Internet, people can communicate and obtain various information, which promotes global progress. At the same time, various information carriers emerge in endlessly. The amount of multimedia data information represented by pictures and videos is growing at an alarming rate with the development of computers. It was once reported that human beings have entered the era of image reading. However, there are so many images containing useful information that people need, and they are scattered in every corner of the world, so the problem is that it is difficult for people to access and obtain useful information. In this context, there is an ur...

AI Technical Summary

Problems solved by technology

[0004] There are two main problems in the existing content-based human skin image classification and retrieval methods. The first problem is the acquisition of human skin images. In the current research, there are many methods commonly used to acquire human skin images, such as : Standard digital cameras, Dermilte Dermoscopy and Proscope HR, although they can collect skin images on the surface of the human body, there are problems in the representation accuracy and accuracy; the second problem lies in the specific algorithm used to retrieve images

At present, the most common algorithms used to retrieve images are the gray-level co-occurrence matrix algorithm and the Tamura texture feature algorithm, but they have a problem, that is, the image retrieval effect for those images whose main objects are prominent and occupy a large area of ​​the image is still poor. Yes, that is to say, their overall image retrieval effect is relatively good, but the key to successful retrieval of human skin lies in details, because the texture of different parts of human skin will present different characteristics, which is the subtle difference in local details, so the two algorithms Content-based image retrieval is satisfactory for image retrieval results where the main object in the image is prominent, but it is not suitable for images that are similar to human skin images that reflect differences in details and subtleties

Images