抽象的

Facial Image Annotation Search using CBIR and Supervised Label Refinement

Tayenjam Aerena, Shobha T.

Image annotation plays an important role in image retrieval and management. Auto face annotation aims to automatically assign a face with the denomination of the corresponding person. However, the face recognition presents conundrum in the field of image analysis and computer vision, and it has received a great deal of attention over the last few years because of its applications in different domains. Mining web facial images on the Internet has emerged as a promising paradigm towards auto face annotation. Therefore, a search assistance tool has been proposed, which helps the face image search based on annotation more efficient. In a face image database which has several face images of many personalities and the labels are not placed properly, which means some images are without label and some are without proper label, in this situation annotation based search will fail. To overcome this problem, an effective supervised label refinement (SLR) approach is proposed which helps in refining the labels of web facial images with human manual refinement effort. To further speed up the proposed scheme, clustering-based algorithm is used. Also the proposed system is enhanced for the two name problem, sometimes a person may have two names and hence confusing the system; this issue is solved in this system. The proposed technique is applied to the automated image-annotation task in our experiments, and hence a demonstration would be made to show that our technique is empirically efficacious and promising for mining web facial images. This technique may be applied in real world applications like social media portals (e.g., Facebook) to automatically annotate users’ uploaded photos to facilitate the search and management of online photo albums