The location method of the main hand-shape feature points

While locating the points used for obtaining the hand-shape features with high class separability, the stability of the point positions is easily influenced by the hand positions and the wearing decorations. This paper improves the linear fitting accurate location method through removing the interference of parts of the experience values. It can reduce the impact of fingers flexibility and improve the accuracy of the finger-tip and finger-root points. In addition, the paper proposes a revised method to locate the wrist point. And the palm length can be used for automatic identification as one of the features in the vector. The experiments can verify the stability of the method through the standard deviation mean. Through the contrast of the matching results between the automatic and artificial measurement, the D-value is 0.7% of the 3-feature vector and 0.4% of 6-feature vector. It can prove the feasibility of the location method used the peg-free images.