Method and system for extracting outlines

Abstract

The invention relates to the technical field of image processing and provides a method and a system for extracting outlines. The method includes: extracting a first image containing an object to be identified from input images; performing affine projection of shape models of automatic storage management (ASM) training sample images onto the first image according to scaling so as to obtain a first shape outline; narrowing and extracting a second image containing the object to be identified from the first image, adjusting a previous shape outline according to scaling so as to obtain a second shape outline, and adjusting the second shape outline according to any feature point on the shape models so as to obtain a third shape outline. Due to the fact that the object to be identified is extracted at least two times, correspondingly the shape models of the ASM training sample images are adjusted automatically at least two times according to the scaling, and the method and the system for extracting outlines guarantee and optimize extraction accuracy, simultaneously avoid problems of poor extraction accuracy and long extraction time when an initial mobile coordinate and scaling coefficients are set in manual mode, and are particularly suitable for extraction of outlines of remote objects to be identified.

Description

technical field [0001] The invention belongs to the technical field of image processing, in particular to a contour extraction method and system. Background technique [0002] In recent years, image recognition technology has been widely used in the fields of identity verification, video surveillance, and human-computer interaction. It mainly includes three steps: target detection, target contour extraction, and target recognition for input images. Among them, target detection refers to the process of judging the position, size or posture of the target in a given input image; target contour extraction is the process of extracting the main features of the target through a certain algorithm on the basis of target detection. Through dimensionality reduction, the amount of data for target recognition is reduced. [0003] Active Shape Model (ASM) is an algorithm that can realize target contour extraction. Specifically, ASM uses a set of discrete feature points to describe the s...

AI Technical Summary

Problems solved by technology

[0006] The purpose of the embodiments of the present invention is to provide a contour extraction method, which aims to solve the problem of manually setting the initial movement coordinates and scaling factors of the shape model on the input image when the existing contour extraction method performs initial contour positioning on the object to be recognized. , resulting in poor extraction accuracy and long extraction time

Images