HAND SHAPE ALIGNMENT-BASED RECOGNITION METHOD

In this research, a simple method is proposed for recognizing hand shape using shape alignment measurement. This proposed method captures colored hand images using simple scanner device, pre-processes them to extract hand shapes, aligns the extracted hand shapes and produces their point correspondences. Then, it uses Mean Alignment Error or MAE criterion to measure similarity/ difference between the aligned hand shapes. Three different simple methods are involved for: finding point correspondences between the aligned hand shapes, ring artifact removal, and ill-defined wrist region correction. In this research, three cases are studied by testing three pairs of hand images. The first pair of images belongs to the same hand, the second pair of images belongs to different hands, and the third pair is formed by images captured from the same hand but in different poses. The MAE of each pair is computed and compared with a predefined threshold value to make the recognition decision. The preliminary results encouraging us to address issues that will improve the used methodology and measurements in the future