Recognition of Chinese finger sign language via gray-level co-occurrence matrix and K-nearest neighbor algorithm

Having a definite expression, Chinese finger sign language obtains vital position in Chinese sign language recognition. In this work, we proposed a method for identifying Chinese finger sign language via gray-level co-occurrence matrix (GLCM) and K-nearest neighbor (KNN) algorithm, which was a vision-based sign language recognition. We captured gesture images and segmented the hand shape, then converted these two-dimensional images into a gray-level co-occurrence matrix and extracted features based on appropriate angle and distance. At the same time, the KNN algorithm was used to classify based on the 10-fold cross-validation mode. We used this model to test 1110 finger images from 30 categories. Finally, the best classification accuracy of 95.3% was achieved. Therefore, our method was considered to be effective for Chinese finger sign language classification.