Method for constructing picture shape contour outline descriptor

Abstract

The present invention discloses a constructing method for the graphic contour descriptor. The method includes the following steps: A. the graphic contour point is detected by point to get contour vector; B. according to the contour vector, the directional distance vector is drawn off; C. according to the directional distance vector, the contour descriptor is constructed. The present invention has the following advantages: firstly, the description of the graphic contour is very intuitionistic, which is similar to the sense of human; secondly, rotation, parallel moving, shrinkage and enlargement can not change the contour, and the contour has good anti-interference ability for shape changing; thirdly, the data structure is very clear, searches can be performed under different resolutions; fourthly, for non-convex shapes, good description ability is also possessed. The contour descriptor constructed by the present invention is effective and has good robustness.

Description

technical field [0001] The invention relates to the field of visual information retrieval, in particular to a method for constructing image shape contour descriptors. Background technique [0002] With the popularity of computers and broadband networks, the promotion of digital acquisition equipment and digital storage equipment, the amount of various multimedia information data including text, images, video, audio, etc. is growing at an alarming rate. How to find useful information from massive data as soon as possible has become a serious problem. Therefore, efficient access to data, especially to visual information data, has become more and more important. [0003] In order to solve this problem effectively, a retrieval method based on visual information content is proposed. This method can automatically obtain the content of visual information through the analysis of its characteristics, thus opening up the way for efficient visual information access. The key problem ...

AI Technical Summary

Problems solved by technology

[0012] Although the feature data obtained by using the method based on the particle radius model is invariant to translation, scaling and rotation, but because the feature data calculates the distance from the contour to the centroid at a fixed angle, the description of the non-convex graphic contour will produce a lot of problems. big error

Especially for radial shapes such as stars, the effect will be poor, because in this case an edge on the contour, regardless of its length, may only be mapped to a point in the feature data, making the method cannot effectively describe the contour of a shape

Images