A hybrid mathematical model for face localization over multi-person images and videos

The reliability and efficiency of face recognition process depends on the effectiveness of facial ROI Extraction. In the surveillance system with multiple object and person images, it becomes more challenging to identify the facial image. In such real time images and videos, the accurate face localization is proposed in this work. This work has presented a hybrid model by using the low level techniques. The window based mathematical filter model is applied along with color model transformation to identify the facial region more accurately. The experimentation is applied on group pictures and videos. The results show the accuracy in facial region extraction.