Graph fourier transform based descriptor for gesture classification

This paper proposes a method for gesture classification based on Graph Fourier transform (GFT) coefficients. GFT coefficients are the projection of image pixel block onto the eigenvectors of a Laplacian matrix. This Laplacian matrix is generated from undirected graph, representing a spatial connectedness between each pixel within an image block. This work proposes a method for generating an undirected graph by using edge information of the image. Edge information of the image is obtained by average sum of absolute difference between the current pixel and its neighboring pixels by using an appropriate threshold. The resulting GFT based feature vector is formed by concatenating GFT coefficients of each block. The resultant feature vector is applied to linear Support Vector Machine (SVM) classifier to predict the gesture class. For NTU and Massey hand gesture datasets, threshold value 30 gives maximum prediction accuracy. We compare the results of the proposed GFT based descriptor approach with Karhunen-Loeve transform (K-LT) and Discrete Cosine transform (DCT) based descriptors on three different gesture datasets: NTU, Cambridge and Massey. Simulation results show that the proposed GFT based descriptor gives a comparable results with Karhunen-Loeve transform (K-LT) and Discrete Cosine transform (DCT) based descriptors for gesture classification.