Gesture Recognition Based On Digital Image Processing

Gesture data is mainly collected from the camera in the system, the system makes feature extraction from the gesture data, makes the machine understand the meaning of gestures.Identification work is made up of three stages: preprocessing, feature extraction, Euclidean distance determination. Noise signal should be removed in the image of pre-processing stage. The color of gestures and background color has different range in the HIS model space, Edge detection operator is used for gesture contour extraction. Fourier descriptors of the contour curve is used for feature extraction. Euclidean distance is used for the gesture recognition. Video streaming through the sample is used for completion of the dynamic gesture recognition.Some statistical experiments were made in this gesture library of 1to10. In the even light. of the background , gesture can be correctly identified basicly.Dynamically gesture recognition can be correctly identified in the even light.of the background.This system has the rotation invariant, proportion scalability translation, the effect of experimenters of rotation invariant, proportion scalability in static gestures library of 1to10 is good.The biggest characteristic of system is accorded with operation of human in data collection addition to traditional wearing gloves and doing color markers on the hand .The method is more close to the operating habit of human. The operator moves more naturally. Deficiencies of the system is vulnerable to external environment such as light, temperature, background color etc. The effection of experiment have relations with the habits. of operator...