CAJ | 학술논문

文章针对一种灰度图像的主基元图模型和提取算法的不足，提出了一种彩色图像的主基元模型和提取算法，该算法是建立结构和纹理的统一表达模型，在遵循最小描述长度的基本原则下，利用匹配追踪（matc-hing pursuit）算法，从给定一幅图像中推理主基元图。实验结果表明，彩色图像主基元图模型比灰度图像主基元在表示上更加完备，能表达在灰度图像主基元图中被“漏检”的主基元。
문장침대일충회도도상적주기원도모형화제취산법적불족，제출료일충채색도상적주기원모형화제취산법，해산법시건립결구화문리적통일표체모형，재준순최소묘술장도적기본원칙하，이용필배추종（matc-hing pursuit）산법，종급정일폭도상중추리주기원도。실험결과표명，채색도상주기원도모형비회도도상주기원재표시상경가완비，능표체재회도도상주기원도중피“루검”적주기원。
By analyzing the limitations of the model and extraction algorithm for primal sketch of gray images ,a modified model and extraction algorithm for primal sketch of color images is introduced . The presented primal sketch unifies the representation of both image structure and texture .Under the minimum description length(MDL) principle ,the primal sketch is inferred from a given input image by using the matching pursuit algorithm . T he experimental results show that the proposed primal sketches of color images are more compact and meaningful than the counterpart of gray images ,and can represent the missing primal sketches in gray images .