Recovery of gray scaled images from contour processed representations

Data compaction can be achieved by parametrizing contours [3] and storing the texture on each side of the contour. Essentially, the image (a two-dimensional structure) is represented by one-dimensional functions. Attneave [1] has shown that most of the information of a scene lies in shapes (thus contours). This paper shows that texture (which complements shape) can also be represented along contours. The paper also discusses the recovery of lost information and suggests two criteria for assessing the correctness of the interpolation method used. The one favored here, an edginess criterion, is used to measure the difference between the original image and the recovered version.