Abstract:
We propose a novel appearance based feature method for face detection using
rigid kernel (template) and its coefficients. The proposed features respond
to pixels of edges of an object(face/non-face) with respect to the kernel as
its coefficients are arranged in a certain order to generate values for better
classification in SVMs (Support Vector Machines). The proposed method
manipulates the symmetric appearance of a face with respect to a rigid
kernel(template).
Description:
This thesis submitted in partial fulfillment of the requirements for the degree of Bachelor of Science in Computer Science and Engineering of East West University, Dhaka, Bangladesh.
