Object recognition using random image-lines

被引：4

作者：

de Vel, O

Aeberhard, S

机构：

[1] Def Sci & Technol Org, Div Informat Technol, Salisbury, SA 5108, Australia

[2] James Cook Univ N Queensland, Dept Comp Sci, Townsville, Qld 4811, Australia

来源：

IMAGE AND VISION COMPUTING | 2000年 / 18卷 / 03期

关键词：

object recognition; line-based algorithm; classification; scale and rotation robustness;

D O I：

10.1016/S0262-8856(99)00016-5

中图分类号：

TP18 [人工智能理论];

学科分类号：

081104 ; 0812 ; 0835 ; 1405 ;

摘要：

View-based recognition is a conceptually simple, relatively robust method for object recognition. Current techniques use small, simplistic object databases requiring, in many cases, large processing training and/or recognition times. In this paper we propose an extension to the view-based object recognition paradigm based on the classification of lines subtending an object of 2D image views that achieves high generalisation recognition rates with reduced computational times. (C) 2000 Elsevier Science B.V. All rights reserved.

引用

页码：193 / 198

页数：6

共 12 条

[1] COMPARATIVE-ANALYSIS OF STATISTICAL PATTERN-RECOGNITION METHODS IN HIGH-DIMENSIONAL SETTINGS [J].