摘要： 针对彩色图像颜色量化过程中整体层次与局部细节之间存在的问题，提出视觉显著性的八叉树颜色量化方法。将视觉注意机制引入图像颜色量化中，分析视觉显著性模型获取视觉显著图方法及理论，根据八叉树颜色量化方法中对细节量化不足的缺陷，结合显著图进行颜色频度统计，增加视觉显著区域的颜色频度。将该方法与八叉树颜色量化图像进行比较验证，可见量化后的颜色具有丰富的层次感，图像细节失真度降低，轮廓清晰。

关键词: 八叉树, 颜色量化, 显著图, 图像处理

Abstract: In order to reduce conflict between whole layer and local detail during color image quantization, this paper proposes an octree color quantization method of visual saliency. It introduces the visual attention mechanism into color image quantization. After analysing the method and theory of getting visual saliency image in visual saliency model, the paper gets the disadvantage that traditional octree color quantization has scarcity of quantizing detail, consults salient image during color register to increase the frequency of color in visual saliency area. Compared with result of octree color quantization, image quantized has more layers in color expression, less distortion in detail and has clear contours.

Key words: octree, color quantization, salient image, image processing

中图分类号: 

• TP391.41