Research of computer image aesthetics’ classification and assessment based on support vector machine

Computer image aesthetics is an interdisciplinary research field, which covering visual arts, psychology, information theory and other disciplines. And it depends on the image processing and computer vision to solve specific problems. This paper will design a comprehensive computer image aesthetic evaluation model, containing aesthetic categories and scores prediction, which will be realized by SVM classifier and SVR algorithm. Experiments show that the model's experimental results are in conformity with human aesthetic perception results. pictorial image through analyzing image's order degree and harmonious degree. Studying the results of other researchers and combining their typical characteristecs, this paper will build a computer image aesthetic evaluaiton model can be widely used in different types. This paper's image aethetic evaluaiton research is mainly composed of three parts, which are aesthetic feature extraction, aesthetic level classification, and aethetic score evaluation. The paper uses machine learning methods to establish the evaluation model. Among them, aesthetic image classification model is based on support vecotr machine, dividing the image to high and low aethetic. Score evaluation model uses the support vector machine to provide specific scores for imange beauty. These two models respectively correspond to the classification and regression problems fo machine learning. It realizes the machine's automatic classification, and provide a score close to people's aethetic habit.