ABSTRACT
Unconstrained face images are known as those images which are unrestricted or those images which are not straightly posed. For identifying a victim or criminal we will take several images of different people which are constant and we can easily find out the victim. But identifying a person with unconstrained face images is difficult task, Example of such images are CCTV cameras. For recognizing a face we have to combine multiple sources of face information known as face media, to identify the person of interest. In this paper the unconstrained face recognitions uses images, videos, sketches and demographic information’s.
In this paper several images and videos will be taken which are not constant, and then the poses are corrected by 3D face modelling algorithm. After the pose corrections we will match the face images to the original image and find how many frames are matched to the original image. Each face…show more content… Automatic face identification techniques are used to identify candidates suspect list .For forensic investigations the available data or media of the suspect may include still face images, video tracks, face sketch, and demographic data. The traditional face matching methods take a single media i.e. only a still face image, video track or face sketch for examination or for analysis report to generate a victim list. In this paper the proposed approach contributes to forensic investigations by taking into account the entire media collection of the victim or criminal face images and then to perform face matching techniques. This way it generates a single candidate suspect list rather than a separate list for each face sample images, thereby reducing the amount of human analysis