基于稀疏非负张量分解的图像分类算法

Study on Image Classification Algorithm Based on Sparseness Non-negative Tensor Factorization

为了降低图像分类算法的计算复杂度,提高图像分类的准确性,本研究提出一种基于稀疏非负张量分解的图像分类算法,首先提取图像本身的结构特征信息得到图像特征数据,再把子空间数据稀疏性作为约束项,添加到非负张量分解目标函数中,再利用稀疏约束的非负张量分解算法对图像数据集进行降维处理,最后使用支持向量机方法对图像数据库进行分类。实验结果表明,本研究提出的算法能有效提高图像分类的准确性并降低计算复杂度。

To reduce the computational complexity of image classification algorithms and improve the accuracy of the image classification, in this study, the image characteristic data was obtained from image structure information, and then the sparsity of sub-space data was taken as a constraint and added in the target function of the non-negative tensor factorization. Next, the sparseness non-negative tensor factorization algorithm was adopted to reduce the dimensions of the image data set. Finally, the SVM was used to classify the image database. Experimental results showed that compared with other algorithms, the algorithm can effectively improve the accuracy of the image classification and reduce the computational complexity.