Research On Algorithm Of Face Iamge Beautification Of Digital Image

In our society, more and more multimedia systems, such as cameras, mobile phones, video conference and so on, have been used, they all need show the digital color image. With the continuous improvement of aesthetic concepts, people are more and more sensitive to the quality of the display images, including the resolution and how they look in pictures. As the growth of the applications of Internet and multimedia technology, people put more attention on variant image processing functions. Currently, there are several package providing functions for beautifying digital face image, such as Photoshop, PhotImpact and so on. Although Photoshop can complete the job in this area, it requires the users to have some knowledge about Photoshop and the job is complex and cumbersome, so we need a new system which can automatically beautify the face image. Recently, there are more and more research on this area. The high-quality image means that the face image is processed clear and beautiful without losing the information is pursued. The algorithm has be proposed according to researching this area, a system for automatically beautifying the face image is completed by processing the images which is JPEG format. They include several steps. Firstly, the face area is detected by using the algorithm of color model. Then for keeping the feature information in the face area without blurring, the improved Niblack binarization algorithm is proposed by detecting the face features, such as eyes, nose, and mouth and so on. Then an improved iterated nonlinear multistage median filter is proposed, it can differentiate the edge and flat regions to remove the undesired components and protect the detail information. Through the above steps, the system can keep the detail information, such as the mouth, nose and so on , while remove the unsatisfactory information, such as spots, wrinkles and other skin factors which affect the aesthetic. Compared with the nonlinear filter system proposed by Arakawa, the above system has the better results.The proposed system has been realized on Visual C++. Experiments on a wide range of sample images prove that this algorithm can effectively eliminate the unsatisfactory factors, such as spots and so on, while keep the image quality and complexity. The system has academic innovation and practical applications.