Research On Hand Shape Recognition System Based On Finger Mid Line Positioning & DSP Implementation

With the development of the social,cyber security has become a serious concern of our life.Traditional identification methods can't meet people's life requirements.Biometric identification technology is gradually replacing traditional methods due to its high security.As an important branch of biometrics,hand-shape recognition technology has been widely used in commercial place.Compared with biometric identification system based on PC,embedded biometric identification system has the advantages of small size,low cost and convenient application,which has a great market.Therefore,a hand-shape recognition system based on finger midline positioning has been designed and transplanted to DSP.The main contents of this paper are as follows:(1)A hand-shaped localization algorithm is designed to solve the problem that the conventional finger midline localization algorithm is susceptible to the irregular outline of finger roots.First,some image preprocessing method are used to remove images noise,then the hand profile is extracted and five fingers are separated,finally,this paper designed a finger midline extraction algorithm based on triangulation.The method does not depend on the location of finger root points and fingertip points,it can effectively avoid the problem of finger midline extraction offset caused by incorrect fingertip positioning and irregular contour near the root point.(2)This paper defines a method for calculating hand shape similarity.This method calculates the similarity by calculating the ratio of the combined area of two finger profiles In order to avoid the effect of unsteady finger outline,the algorithm chooses up 90% of the finger outline to calculate the similarity.Compared with the existing contour-based hand shape recognition algorithm,this algorithm has better stability,higher recognition rate,recognition rate is 99.89%.(3)The hardware and software structure of the hand-shape recognition system based on DSP are designed.The algorithm proposed in this paper is verified on DSP.