This preview shows pages 1–2. Sign up to view the full content.
This preview has intentionally blurred sections. Sign up to view the full version.View Full Document
Unformatted text preview: Detection of human faces in colour images C.Chen S. - P. C h i a n g Indexing terms: Video signals, Image recognition, Human faces Abstract: In modern multimedia systems, video and image signals usually need to be indexed or retrieved according to their contents. Colour characteristics are proposed for use in detection of human faces in colour images with complex backgrounds. The proposed face detection method first uses a neural network to classify the images and then segments the candidate face regions. Then, an energy thresholding method which can take the shape, colour and edge characteristics of the face features into the extraction process is devised to extract the lips. Finally, three shape descriptors of the lip feature are used to further verify the existence of the face in the candidate face regions. The experimental results show that this method can detect faces in the images from different sources in an accurate and efficient manner. Since faces are common elements in video and image signals, the proposed face detection method is an advance towards the goal of content-based video and image indexing and retrieval. 1 Introduction Video and image signals are very important elements in multimedia systems. Ideally, the multimedia system should provide users with the ability to query and edit video and image signals conveniently. To accomplish this goal, the video and image signals have to be manipulated according to their contents [l]. In [l], Smoliar and Zhang described this kind of video manip- ulation as content-based video indexing and retrieval. Before the video and image signals can be stored or retrieved according to their contents, the objects in the signals have to be recognised. Human faces are key elements in image and video signals. The detection of human faces in the video pic- ture is therefore an important step toward the goal of content-based video indexing and retrieval. The diffi- culty in face detection is that there is a large number of image frames in a common video signal. In addition, the background of the image is usually complex. The two basic approaches to face recognition, as reviewed in , model-based vision techniques (e.g. the Hough 0 IEE, 1997 IEE Proceedings online no. 19971414 Paper first received 4th October 1996 and in revised form 2nd June 1997 The authors are with the Department of Information Engineering, Feng Chia University, Taichung, Taiwan 407, Republic of Chna transform  and template matching ), and feature extraction and location (e.g. eyes  and lips ) are usually very time consuming and can only be applied to images with simple backgrounds. Recently, Yang and Huang have developed some simple rules based on the contrast between the face and background to locate candidate areas of the faces . However, their method is still time consuming. It takes about 60-120s on a SUN-4 station to locate the face in a 512 x 512 picture and only 83% of faces are detected....
View Full Document
This note was uploaded on 10/23/2010 for the course COMMINUCAT 123 taught by Professor Ali during the Spring '10 term at Masaryk University.
- Spring '10