文章提出了一种基于L1-L2混合噪声模型(HNM)方法用于图像、视频的超分辨重建,该方法有效利用L1范数在突变区域较好保持图像边缘信息的特点,同时兼顾了L2范数对图像平坦区域的噪声抑制特性.针对求解过程中噪声分布的改变,L1、L2范数的选择问题,提出一种自适应隶属度(AMD)方法,该方法不仅能达到理想的实验效果,而且大大地减少了迭代次数与运算时间.%L1-L2 hybrid noise model (HNM) method is proposed in this paper for image/video super-resolution. This method has the advantages of both L1 norm minimization (i.e. edge preservation) and L2 norm minimization (i.e. smoothing characterization). In view of noise distribution changing and selecting L1 norm minimization or L2 norm minimization, the paper propose an efficient adaptive membership degree (AMD) method, which get the ideal result but the proposed AMD method can reduce the number of iterations and save much computational cost.
展开▼
