Image processing method and system based on entropy regular fuzzy non-negative matrix factorization

The invention discloses an image processing method and system based on entropy regular fuzzy non-negative matrix factorization. The image processing method based on entropy regular fuzzy non-negative matrix factorization comprises the following steps: acquiring an image matrix to be processed; decomposing the image matrix into a product of two matrixes by adopting an entropy-based regular fuzzy non-negative matrix decomposition model; and obtaining an image processing result based on the two matrixes obtained through decomposition. According to the image processing method provided by the invention, the technical problems of low precision, poor interpretability and the like when an existing traditional non-negative matrix factorization model is used for processing image clustering or hyperspectral unmixing can be solved. 本发明公开了一种基于熵正则模糊非负矩阵分解的图像处理方法及系统，所述基于熵正则模糊非负矩阵分解的图像处理方法包括以下步骤：获取待处理的图像矩阵；采用基于熵正则模糊非负矩阵分解模型将所述图像矩阵分解为两个矩阵的乘积；基于分解获得的两个矩阵，获得图像处理结果。本发明提供的图像处理方法，可解决现有传统非负矩阵分解模型处理图像聚类或高光谱解混时存在的精度低、可解释性差等技术问题。