Method and device for identifying human skin color region

Abstract

The invention discloses a method and a device for identifying a human skin color region. By the method and the device, the skin color region can be quickly and accurately obtained, and accuracy and efficiency of picture filtration are improved. The method for identifying the human skin color region comprises the following steps of: acquiring hue-saturation-value (HSV) data of pixel points in an image to be detected, and clustering the pixel points according to the HSV data to obtain a clustering result image; based on the statistical result, selecting the threshold range of the hue (H); performing region segmentation according to H components of the pixel points in the clustering result image and the threshold range of the H to obtain a skin color region of the image to be detected; dividing the pixel points in the skin color region by using an overflow filling algorithm, wherein the pixel points satisfying the communication condition are divided into the same skin color blocks; and performing feature extraction on the skin color blocks respectively, and executing filtration of the image to be detected.

Description

technical field [0001] The invention relates to the technical field of image recognition, in particular to a method and device for recognizing human skin color regions. Background technique [0002] Skin color area segmentation technology is one of the hotspots in current image segmentation research. It has important applications in face recognition, expression recognition, human hand tracking, human-computer interaction, moving human target tracking, and pornographic image filtering. At present, the proliferation of pornographic pictures on the Internet has caused great harm to minors. Whether skin color segmentation can be quickly and effectively realized has an important impact on the accurate identification of pornographic pictures. [0003] In the existing filtering scheme of yellow pictures, the whole human body is found on the yellow picture through the segmentation of the skin color area first, and then the movement is analyzed after the human body is found, and some...

AI Technical Summary

Problems solved by technology

[0004] However, there are still many problems in the existing filtering schemes. For example, the existing schemes cannot accurately segment the skin color area. Usually, the action recognition is based on the feature extraction of the skin color area of ​​the whole human body, which makes feature extraction difficult. Filtration is less accurate

Images