画像解釈言語PILSの文法

概要

論文の詳細を見る
This report is intended to serve as a concise, ultimate reference for both programmers and implementers. PILS is a problem oriented language designed for image representation and understanding of such silhouette images as stomach barium-filled X-ray image, micro-scopic image, chinese chracter etc. 1) PILS represents an image by two data structures; macro-contour and micro-contour. 2) Macro-contour is composed of five hierachical picture elements, which are represented by twelve attributes : area, height, width, etc. Micro-contour is represented by an arc sequence. An arc is represented by seventeen attributes : length, tangent, shape, etc. 3) The attribute and relation of picture element are accessed by picture operators. 4) Searching statement and associative memory are arranged to describe pattern recognition processes. 5) PILS user can define his own picture operators.
山梨大学の論文