An improvised methodfor detecting face as ROI in video

Abstract

Objective of the paper is to detect human faces in a video. Lenient decision-based face detection structure is developed. Strength of this structurecan be calculated for a variety of poses, complex backgrounds, face expressions, existence or nonexistence of structural components, orientation of image, condition of imageand blur noises. In particular paper will focus on acolor-based skin model for finding face inside a video which is a unique Region of Interest (ROI).This ROI is calculated, with Improvised Voila Jones Algorithm using Adaboost Cascades. The algorithm is capable of detecting range of unlike faces along with difficulty associated to different illumination conditions, different facial expressions, Image conditions and complex background details. The proposed algorithm detects faces with minimum calculation time while maintaining high detection precision.Then the unique ROI is used for embedding secret information to improve security, robustness, flexibility and imperceptibility of stego file.