Image Recognition Based on Sparse Enhancement and Collaborative Filtering Algorithm

摘要

With the development of compressed sensing theory, a large number of image denoising, image fusion, image super-resolution and collaborative filtering algorithms based on sparse enhancement have been proposed, and they have been widely used. However, the current research on sparse enhancement and collaborative filtering algorithms that have been applied to various image recognition fields still needs improvement. In this regard, the purpose of this article is to study image recognition based on sparse enhancement and collaborative filtering algorithms. In this paper, the score prediction value of the traditional collaborative filtering algorithm is affected by the number of common items. It starts by reducing the sparsity of common scoring items in the closest neighbors, and includes the closest neighbor users who are similar to the target user but have not yet scored the target item. Finally, the global and local features of the image are extracted to construct global and local feature dictionaries. The feature vectors of the image are jointly coded on the corresponding feature dictionaries. In addition, the size of the contribution of different features is reflected in the coding domain through adaptive weighting. The classification of the image will be recognized, which is determined by the overall coding error. The experimental results prove that the sparse representation and collaborative filtering algorithms are highly feasible. Both occluded and unoccluded image recognition experiments performed very well, and the recognition rate reached more than 90%. It is very helpful for the development of image recognition system in actual industrial production.