基于视觉的目标提取方法综述

Review of Vision-Based Object Detection Methods

感兴趣目标提取是计算机视觉领域一个经典和基础的研究课题,是正确实现几乎所有高层次视觉任务的关键.由于实际采集环境中存在许多不可控制的因素,使得快速准确的目标提取成为几十年来研究者一直面临的挑战.描述了七类代表性的目标提取方法及其优缺点,并通过二次分类分析了它们的适用范围和理论脉络.

Object extraction is a classical and fundamental problem in computer vision field, and the key to solve almost all the high-level vision tasks. Due to many uncontrolled factors in real environments, it becomes a challenge that researchers have faced in several decades to extract the objects of interest accurately and quickly. Seven representative types of object extraction methods along with their advantages and disadvantages are described in this work. Second classification is made to analyze their range of application and theoretical contexts.