Recovery of Edge Pixels for Facial Expression Recognition using Edge Fusion Approach

Detection of edges is an inevitable part in the realm of Image Processing. In detecting the emotional state of a person, the facial feature extraction of the facial components like eyes, eyebrows, lips etc. play a significant part in accurately classifying the facial expressions wherein the role played by edge detection is very vital. The discontinuity or abrupt change in the pixel intensity is typically represented by edges. Although there are numerous edge detection operators, getting the finest outcomes from the current system is a huge difficulty. Perfect edge detection ensures capturing of accurate features. This paper presents an edge fusion technique for facial images which combines the Sobel and the Canny detected edges of the facial images with the Active Appearance Model (AAM) which is also a statistical face modelling algorithm so that the outcome of the fusion is used to visualize the completeness of the edges and the localization of the facial components in an enhanced way. The performance of this edge fusion technique discussed here seems to be promising in deducing the image metrics necessary to confirm the emotional condition depicted by the face.