Font Size: a A A

Understanding Of Visual Art And Style Of The Classification Based On Sparse Coding

Posted on:2013-09-28Degree:MasterType:Thesis
Country:ChinaCandidate:C L WanFull Text:PDF
GTID:2248330374459653Subject:Signal and Information Processing
Abstract/Summary:PDF Full Text Request
Visual system is more perfect than any other artificial digital image signal processing system, in order to improve digital image processing technology, bionics is a good way. Sparse coding as an artificial neural network method can model the receptive fields of simple cells in the mammalian primary visual cortex (area V1) in brain, and can effectively extract the internal features of the image. When using the Sparse coding algorithm for the visual art works’style understanding and style classification, on the one hand, can well combine the computer graphics, biology, art and other disciplines, and can further promote their development; on the other hand, visual art works’ style classification can help the visual art appreciators and Reviewers for understanding and researching the visual arts’ manifestations, and further enhance the artistic quality.This paper mainly focus on the sparse coding theory, and it’s application in the field of visual art works’ style understanding and style classification. The main work on the following aspects.First of all, deeply study of sparse coding algorithms, including the algorithm model, the learning process and the algorithm pre-pretreatment work.Secondly, basis on fully understanding the theory of sparse coding algorithm, then use this algorithm for different style visual art works’ coding, getting the features of basis function and sparse coefficient. By deeply analysis and research these features, in order to achieve visual art works’ style understanding and style classification..Thirdly, basis on the features of basis functions and sparse coefficient showed a larger inter-class distance and small within-class distance,we use the fourth-order statistics (kurtosis) for the higher-order statistics of the image data, and then use the mathematical statistics for the visual art works’style understanding and style classification.The experimental data prove that sparse coding algorithms can feasibility achieve the visual art works’ style understanding and style classification.Finally, we summarizes and discuss the shortcomings and the problems in our experiment, and outlook the further research directions.Studies have shown that the sparse coding algorithm is a good way and can get good experimental results for the visual art works’ style understanding and style classification.
Keywords/Search Tags:Sparse coding, Visual arts style, Image understanding, Styleclassification, Basis functions, Sparse coefficient
PDF Full Text Request
Related items