Noise Removal and Face Age Prediction from CCTV Video Feeds to Aid Digital Forensics

Abstract

The analysis and computation of a person’s body measurements are attained by anthropometry. This examination is done with a perspective of the significance of anthropometric files of the facial feature in surgery, forensic medication and medical designing. The principle objective of this exploration is enhancement of face component landmarks by setting up a numerical relationship among face feature elements and utilized feature points that are relevant for age prediction. Since chosen face component focuses are situated to the range of mouth, nose, eyes and eyebrows on face pictures, all fascinating face element focuses are derived precisely. In order to experiment with this introduced technique; 16 Euclidean distances are computed from 18 chosen points of face feature landmarks vertically horizontally. The mathematical relation between horizontal distances and that in vertical direction are initiated. Furthermore, due to age progression it is further determined that the distances of the face landmarks exhibit fixed ratios. The distances among the described features points gained with reference to the progression of age of a person from one’s childhood yet the proportion does not vary of the distance(δ=1.618).For Experimentation, first face is recognised from the video and age is predicted using the anthropometric features. The video is placed as video frames. Video frames is given as input for age estimation. For Face Age prediction SVM and CNN is used for experimentation purpose. Age Estimation is done from the video frames. Noise is removed from the video frames and then age is estimated from the frames.