稀疏矩阵文档图像处理

Document Image Processing Based on Sparse Matrix Representation

讨论基于稀疏矩阵的文档图像存储及处理方法 .采用三向量法或链表法表示稀疏图像 ,然后在稀疏域直接实现某些基于临域运算的图像处理算法 .分析表明 ,对于具有显著特征的文档图像能有效地节省存储空间并提高计算效率 .以卷积运算和一种文档图像处理运算为例 。

Storage and processing of document images based on sparse matrices are discussed. The sparse image is first converted into a more compact form consisting of either three vectors or a linked list. Image processing operations featuring neighborhood computations are performed directly in the sparse matrix domain. For images that are sufficiently sparse, the described method can effectively save storage as well as processing time. Finally, sparse matrix domain implementations of 2D convolution and a noise removal algorithm are presented.