Facial image retrieval, identification, and inference system

Abstract

Recognition of a human face is very easy even to a child, but is extremely difficult for computers. Here we present a Computer Aided Facial Image Identification, Retrieval and Inference System (CAFIIRIS) for criminal identification. The system stores and manages facial images and criminal records, providing necessary image and text processing, and editing tools. Inference of facial images of different ages of a person is also possible. Access to facial images can be done via key words, fuzzy descriptions, and visual browsing. A facial image database system stores and manages a large amount of facial images together with text-based criminal record. It provides users with a flexible means to manipulate, archive, retrieve, and make use of facial images and text data. The facial images are visual rather than descriptive. Each digital image is a large array of pixels of various sizes, and a facial image database contains thousands or even hundred thousands of images. Therefore, this huge visual database needs special techniques for its management, namely, embedded functions for image pre-processing, feature extraction, presentation (screen display and report formatter); visual access to image data via special indexing techniques; application-specific image inference to derive new images based on images and other available information.