Author : Miss. Apurwa V. Atre and Dr. Mrs. R. A. Khan

Face recognition has increased a significant situation among most usually utilized utilizations of image processing. With the fast development in multimedia contents, among such content face recognition has got much attention especially in past few years. Face as an object consists of distinct capabilities for detection; therefore, it remains maximum challenging research area for scholars within the field of computer vision and image processing. Probe face pictures are delivered in an unconstrained environment. A face might be impeded by shades, a cap and a scarf, captured in various poses, situated halfway out of cameras eld of view. Human face plays a significant job in our social collaboration, passing on individuals’ personality yet it is a powerful item and has a high level of inconstancy in its appearances. The issue of perceiving a discretionary fix of a face picture remains to a great extent unsolved. This investigation proposes another incomplete face acknowledgment approach, called Dynamic Feature Matching, which consolidates Fully Convolutional Networks, Principle Component Analysis and Sparse Representation Classication to address halfway face acknowledgment issue paying little heed to different face sizes. DFM doesn’t require earlier position data of partial faces against an all encompassing face.

Keywords: Dynamic feature matching, Partial face recognition, Gabor filter, Principle component Analysis, Fully convolutional network.