A natural approach in image processing and pattern recognition: Rotating neighbourhood technique, self-adapting threshold, segmentation and shape recognition

摘要

Sentence description of objects are utilized in the model. The description is transformable into graphs. Fast shape coding is accomplished by an operator translated along linear objects or contours. A separate node of a graph is obtained for each new position of the operator. After a reduction, the graph is equivalent to a sentence describing an object. A technique incorporating a rotating neighbourhood is introduced and utilized for image enhancement and texture derivation. The technique preserves edges of objects. A threshold which adapts itself to an unknown noise level and image segmentation into individual objects are also presented. All the methods discussed satisfy the basic condition on representing images.