...
首页> 外文期刊>Applied mathematics and computation >Is there a universal image generator?
【24h】

Is there a universal image generator?

机译:是否有通用图像生成器?

获取原文
获取原文并翻译 | 示例
           

摘要

Synthetic pattern generation procedures have various applications, and a number of approaches (fractals, L-systems, etc.) have been devised. A fundamental underlying question is: will new pattern generation algorithms continue to be invented, or is there some "universal" algorithm that can generate all (and only) the perceptually distinguishable images, or even all members of a restricted class of patterns such as logos or letterforms? In fact there are many complete algorithms that can generate all possible images, but most images are random and not perceptually distinguishable. Counting arguments show that the percentage of distinguishable images that will be generated by such complete algorithms is vanishingly small. In this paper we observe that perceptually distinguishable images are compressible. Using this observation it is evident that algorithmic complexity provides an appropriate framework for discussing the question of a universal image generator. We propose a natural thesis for describing perceptually distinguishable images and argue its validity. Based on it, we show that there is no program that generates all (and only) these images. Although this is an abstract result, it may have importance for graphics and other fields that deal with compressible signals. In essence, new representations and pattern generation algorithms will continue to be developed; there is no feasible "super algorithm" that is capable of all things.
机译:合成图案生成过程具有各种应用,并且已经设计了许多方法(分形,L系统等)。一个基本的根本问题是:是否会继续发明新的模式生成算法,或者是否存在一些“通用”算法可以生成所有(且仅此)在感知上可区分的图像,甚至是受限类型的模式(例如徽标)的所有成员?或字母形式?实际上,有许多完整的算法可以生成所有可能的图像,但是大多数图像是随机的,在感知上无法区分。计数论点表明,由这种完整算法生成的可区分图像的百分比将逐渐减少。在本文中,我们观察到可感知区分的图像是可压缩的。使用该观察结果,显然算法复杂度为讨论通用图像生成器的问题提供了适当的框架。我们提出了一个自然的论文来描述可感知的可分辨图像,并论证其有效性。基于此,我们表明没有生成所有(且仅)这些图像的程序。尽管这是一个抽象的结果,但对于处理可压缩信号的图形和其他领域可能很重要。从本质上讲,将继续开发新的表示和模式生成算法。没有可行的“超级算法”能够处理所有事情。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号