Aesthetic Assessment For Digital Portrait Photos

In recent years, digital intelligent products become a major bright spot digital industry, intelligent digital products favored by consumers, whether domestic brand or foreign brands are in the focus on the development of intelligent products. In addition, with the increase of the popularity of mobile phone and digital camera shooting function, it is leading to explosive growth in the number of digital photos, a variety of photography exhibitions and contests are fully launched all over the country. first of all, in carrying out these tasks,we are faced with the problem of a large photos of audition. It is not only obviously subjective that basing on photographers or viewers’own aesthetic ideas to evaluate the photos in traditional evaluation of digital photo quality, but also it is a inefficient and all-consuming task as the needle-in-the-haystack choices, in this way, it is difficult to ensure consistency inspection standards. Therefore, in daily life, we can use advanced computer technology to make analysis aesthetic evaluation of various photos, it not only can save time, but also achieve fairness and justice. In addition, the study involves a number of innovative frontier subjects, for example, aesthetic art, aesthetics of photography, computer science etc. It is very important to the value and application prospect.Supported by graphic image processing technology and photography aesthetics, this essay combines the features and aesthetic elements of digital portraits. Then it extracts faces in digital portraits using skin color, grid analysis and SMQT and SNOW classifier, based on which five features representing the quality of digital portraits are proposed. Applying these features to a same group of photos and comparing the results with human scoring experiment, the author finds that the quality assessment by computers in a way differs from human scoring. However, scores assessed by computers tends to be more stable with clearer indicators, not affected by subjectivity. The main content of this essay includes the following aspects.Firstly, the principal theory relevant to this essay will be introduced and summarized, starting with a brief summary of computer aesthetic theory. Then the fundamental photography theory will be explained, including focal length, exposure, depth of field and white balance etc. influences exerted by these aspects will also be illustrated. Next the basic elements of digital portraits will be explicated, including image structure, Manhattan distance and facial expressions etc. which affects people’s aesthetic point of view on digital portraits.Additionally, features to assess the quality of digital portraits will be extracted. Three approaches will be applied to face segmentation of digital portraits, based on which Hough transformation and K-mean clustering method will be used to extract the contour of eyes. Then, features such as expression evaluation and the proportion of face in portraits will be extracted. The influence of different features on the assessment of aesthetic perceptions of photos will be analyzed through experiment.Lastly, according to the features extracted, the standard of scoring by computers on aesthetics of digital portraits will be proposed. A certain number of photos will be selected, among which single-face photo and multi-face photo are included. The computer will mark these photos. Additionally, these photos will be given to different people to review and mark, out of which an average score will be gained. Then merits and demerits of computer scoring and human scoring will be compared respectively.The experimental result indicates that it is effective to apply Matlab to the quality assessment of digital portraits. Although this approach in a way differs from subjective human scoring, scores evaluated by computers proves to be more stable with clearer indicators, not affected by subjectivity. This approach has a wide range of use and it should be applied to photography or photography competition, which will help photographers and staff selects high quality photos.