论文部分内容阅读
分析比较了基于方差最小标准和基于最大类内距离最小标准的彩色图像量化算法．方差最小算法着重于最小化整幅图像量化前后的总方差；最大类内距离最小算法着重于减少类内任意点与其代表点的最大距离．文中提出了一种结合两种标准的新算法，在限制类内点与其类代表点最大距离的同时，尽量降低整幅图像量化前后的总误差．经验证，该方法有较好的量化效果
The color image quantization algorithm based on the minimum variance standard and the minimum criterion based on the maximum intra-class distance is analyzed and compared. The minimum variance algorithm focuses on minimizing the total variance before and after the entire image is quantized. The maximum distance algorithm within the largest class focuses on reducing the maximum distance between any point in the class and its representative point. In this paper, a new algorithm combining two criteria is proposed, which minimizes the total error before and after quantification of the entire image while limiting the maximum distance between the class points and their representative points. It is verified that this method has better quantification effect