| Hi-index | 0.00 |
The problem of recognising fingerspelling alphabets for sign language interpretation is addressed in this paper. An appearance based model is proposed. The proposed model suggests a modification to the existing diagonal FLD model at two stages, one at rearranging of images and the other at adjusting the contrast of the images by the use of histogram equalisation. The proposed model is called SecDia FLD. An extensive experimentation conducted on a large fingerspelling dataset revealed the superiority of the proposed model. In addition, we have also brought out the effectiveness of the proposed model on the Yale face dataset