Local shape matching method based on outline random sampling

Abstract

The invention discloses a local shape matching method based on outline random sampling. The local shape matching method based on outline random sampling comprises the steps that all sub-outline sequences of a target outline to be matched and random sub-outlines on a template outline are extracted through a geometric round radius method; the length and width ratios of the minimum rotation bounding rectangles and the outline area features of a target sub-outline and the random sub-outlines of the template are extracted, and primary matching is conducted; mean value distance chain code and mean value angle chain code features, based on angle classification, of the sub-outlines of the outline points obtained after primary matching are extracted; secondary matching is conducted between the chain code features obtained through the random sub-outlines of the template and the sub-outlines obtained after primary matching, so that a small number of optimal matched outlines are obtained; a coordinate conversation matrix of the random sub-outlines of the template and the matched sub-outlines are calculated, and the outline of the whole template is projected to the target outline through the conversation matrix; the coordinate conversation matrix of the sub-outlines is updated; and the optimal matched outline is recognized. By the adoption of the local shape matching method, a matched result can be rapidly and reliably provided, the image registration calculation efficiency is high, and the adaptability is high.

Description

technical field [0001] The invention relates to a shape matching method, in particular to a local shape matching method based on contour random sampling. Background technique [0002] Vision is the main means for human beings to observe and recognize the world. Most of the information that human beings obtain from the outside world comes from the visual system. When observing the surrounding environment, people first notice the color, texture, shape and spatial relationship of the object and the surrounding environment, among which shape is one of the most basic characteristics of the object in the sense of actual perception. Shape matching refers to measuring the similarity between shapes according to certain measurement criteria. It is one of the basic problems of automatic image recognition and understanding, and has a wide range of applications in computer vision, pattern recognition, remote sensing image analysis, text recognition and other fields. However, traditional...

AI Technical Summary

Problems solved by technology

However, traditional shape matching methods have low registration rate, slow registration speed and poor adaptability

Images