Detecting Smile and Forgery in Digital Color Images

• Main Authors: Roya M.m Elrefaei, 羅雅
• Other Authors: 貝蘇章
• Format: Others
• Language: en_US
• Published: 2009
• Online Access: http://ndltd.ncl.edu.tw/handle/64467847089017165412

碩士 === 國立臺灣大學 === 電信工程學研究所 === 97 === This thesis proposes three techniques for digital image processing, in which we designed algorithms, enhanced techniques and came up with good results. There are open literature dealt with detecting human face expressions and most of them are done using neural network technique that means very time consuming. First chapter proposes a new simple and an efficient algorithm for smile detection in frontal face; in which it could be exploited as one unique function of camera. In our approach, the smile detection technique consists mainly of three steps used on segmented faces: the first one is by enhancing the lamination in shadow areas of the skin image, afterwards, localizing technique for tracing facial features is done to find the mouth area, and finally the smile is detected using our proposed technique. The second chapter proposes a new method for detecting forgery in digital color images, since there are many efficient ways to create these images; there are also variety tools to detect them where each image is a unique challenge and need different tools to detect them. In most of the artificial images, it is very hard to ensure that all objects are consistently illuminated and therefore the shadow is consistently distributed; from this point, we developed a method which discovers the shadow inconsistency among objects in digital forgery images, our algorithm is done by detecting skin regions from user segmented areas, then shadow and highlight skin regions are found, therefore, vectors are drawn from the center of gravity of the skin highlight region to the center of gravity of the skin shadow region in each skin segmented areas. Face detection is considered as a case from object-class detection, in which the task is to find the locations and sizes of all objects in an image that belong to a given class, such as peoples, building, and cars. Face detection is not only used as a part of a facial recognition system, but also used in video surveillance, human computer interface and image database management, furthermore, some recent digital cameras use face detection for auto-focusing task. At the third chapter, we introduced simple face detection technique, which is an algorithm to determine the locations of human faces in digital images and ignoring anything else, it is based on segmenting the input image into skin and non-skin regions, then by using facial feature, we decide whether this skin region would be a part of the face or not.