Hand Detection Based On Skin Color And Block Feature

With the development of the science and technology and the advancement of the times,the hand recognition methods have got greatly improved.Hand recognition is to mark the hands in the dynamic image files or static images.Hand recognition has a broad prospect in the fields of recognition of sign language,body feeling game,dialogue between people and computers and other fields.At the same time,hand recognition can also serve as an important basis for human body detection in the disaster spots.But human joints are many and irregularlead to that the recognition is difficult.Hand recognition in many research subjects has a certain challenge.This thesis does some researches on the hand recognition focusing on static images and this thesis will provide a hand detection algorithm with a good performance for the hand recognition methods through studies.This paper tries to use the characteristics of the regional articulations of the hands to identify the hands.The characteristics of the regional articulations of the hands is similar to that of an algorithm of Haar features,whose feature values reflect the gray level changes of the pixel of the various points and we can get a feature classifier through training and learning for the blocks from small to big in picture chart and conduct continuous changes for the mistakes or errors made by the feature classifier to construct an accurate training classifier for the features of the blocks.Adopt complexion segmentation and characteristics of hand articulation algorithm to identify the hand recognition.First of all,complexion is a unique human characteristic which does not change with the shapes and sizes,and is quite different from colors of other objects and is an important mean for people to be different from other things.And the skin feature extraction algorithm needs a small amount of calculations with fast execution speed and good aging performance.But the light has a large influence on the complexion segmentation and light treatment is used before the identification for the complexion area.Then we adopt complexion segmentation algorithm to conduct segmentation for human skin area and the non human skin area in the static images,select the complexion area of the human body and make preparations for utilizing the characteristics of hand articulation for hand recognition.Secondly,use characteristics of hand articulation to detect again for the human skin color area,if the pictures contain the hand information and mark the area where the hands locate.Then this method utilizes the human skin color information and characteristics of hand articulation with high speed detection and high accuracy and it achieves more accurate in human hand detection.