自适应视觉基元图象编码

An Adaptive Visual Primitive Coding of Images

本文给出一种自适应视觉基元图象编码方法(Adaptive Visual Primitive Image Coding),它利用了现有的人类视觉系统模型及研究结果,采用一系列独立于自然图象统计特性而与人类视神经元特性紧密相关的视觉基元,对图象进行分割、匹配运算,进而对与自然图象局部区域最佳匹配的视觉基元进行编码,具有运算简单、实现方便的优点.计算机模拟实验表明,在较高数据压缩比(21:1～30:1)时,解码图象的主观质量令人满意.

An adaptive visual primitive image coding method is presented in this paper. Based on the existed human visiual system model, a series of visual primitives, which are designed independent of the statistical properties of natural image but dependent on the properties of human visual neuroes, are used to segment and match the image to be coded. The visual primitives which are best matching with the local areas of natural image are then coded. This method is simple in computation and easy to realization. It is found that subjectively more pleasing decoded image is obtained at relative high compression rates at 21:1 to 30:1.