Image Synthesis into a Procedural Format

将对直接将图像合成为程序格式进行研究。将要开发的现实成像技术将确定程序描述所需的参数和系数，而不是寻找组成空间表示的单个像素。这种方法将允许新算法利用仅计算空间形式时不可用的信息。这应该可以利用图像的统计特性和视觉掩蔽的效果。通过使图像合成技术的输出更接近由图像压缩算法产生的输出，将避免计算不必要的信息，并且最终结果将更符合当前存储或传输计算机图形图像的实践。最后，使用基于感知的程序表示将有助于使用视觉度量来控制如何计算图片并决定其是否具有足够的质量。***

The synthesis of images directly into a procedural format will be investigated. The realistic imaging techniques to be developed will determine the parameters and coefficients necessary for a procedural description instead of finding the individual pixels that compose a spatial representation. This approach will permit the new algorithms to take advantage of information that is not available when only the spatial form has been computed. This should make it possible to exploit such things as the statistical properties of images and the effect of visual masking. By moving the output of an image synthesis technique closer to what is produced by an image compression algorithm, the computation of unnecessary information will be avoided and the final result will be more consistent with current practices for storing or transmitting computer graphic images. Finally, the use of perceptually based procedural representations will facilitate the use of visual metrics to control how a picture is computed and to decide whether it is of adequate quality. ***