|
|
【答案】应助回帖
商家已经主动声明此回帖可能含有宣传内容
★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ... yuxintian: 金币+70, 翻译EPI+1, ★★★★★最佳答案, 太感谢了,确实翻译得很好 2015-10-10 05:22:13
Jacquin在 他的论文中提出,一个图像 不应该认为 是 整个图像的 变形拷贝和粘贴,而应该是 图像的 小部分的 变形拷贝和粘贴。例如,云的 一部分 当然不会像 带有云的 整个风景,但 却可能 发现云 或其它结构的 另一小部分 看起来像云,由此打破了 全局迭代函数 映射 的 僵硬观点,为 图像压缩研究 带来了 新的方向。他实现了 第一个 自动分形压缩 系统,如图4.1所示,首先 把图像 划分成 固定大小的方块,然后对 每个小块 寻找匹配的 图像区域。这种 构造 被称为 局部的 或分区的 迭代函数系统。
Jacquin proposed in this paper that a image should not be regarded as the anamorphic copy and paste version of the whole image,but should be regarded as the anamorphic copy and paste version of a small part of whole image. For example, a small part of cloud is surely not like the whole scenery where the cloud is included, but it may probably find that a small part of cloud or other structure looks like the cloud. Therefore the opinion of global iteration function mapping was broken, charting a new approach for image compression study. He is the first one who realizes the automatic fractal compression system shown in Fig.4.1. First, the image is divided into blocks in a constant size, the each block seeks the matched image region. Such structure is called the regional or divisional IFS.
为 每一个小块 寻找 匹配的 图像区域,实际上 就是寻找 该小块的 局部迭代函数系统或 生成元,把 该图像区域 称为 定义域块(domain块) ,而 把要匹配的 每个小块称 为 值域块(range块) 。一般 所选定义域块 的边长 为值域块边长 的两倍大小,要把它 收缩到 和 值域块 一样大小后 才可以 匹配值域块。由前面所述,生成元 映射形 如 , 其中 a为压缩系数, b为偏移量,c 为收缩后的 定义域块,称为 代码本块,d 为维数与 一致的 单位向量。匹配过程 就是寻找 c,使之满足方程1 。
Actually, seeking the matched image region for each block is to reach the local iteration function system or the generating element for each block. The image region is called domain block, while each block is called range block. Normally, the side length of domain block is twice larger than that of range block, the two blocks can be matched with each other after their side lengths are equal. Based on former analysis, the generating element mapping is like..., wherein, a is the compression coefficient, b is offset, c is the domain block after compression, which is called codebook block, d is the unit vector with the same dimensions as that of....The matching process is actually the process of seeking c, making it meet the equation 1. |
|