中国科学院沈阳自动化研究所机构知识库
Advanced  
SIA OpenIR  > 光电信息技术研究室  > 学位论文
题名: 基于视觉注意的小目标检测方法研究
其他题名: A Small Targets Detection Method Based on Visual Attention
作者: 谢文亮
导师: 朱丹
分类号: TN911.73
关键词: 视觉注意 ; 小目标检测 ; 显著图 ; 图像分割 ; FPGA
索取号: TN911.73/X54/2012
学位专业: 模式识别与智能系统
学位类别: 硕士
答辩日期: 2012-05-28
授予单位: 中国科学院沈阳自动化研究所
学位授予地点: 中国科学院沈阳自动化研究所
作者部门: 光电信息技术研究室
中文摘要: 随着信息技术的迅速发展,作为信息载体之一的图像发挥着愈来愈大的作用。它在为人们提供巨大信息量的同时,也给手工加工和处理带来了困难。在多种多样的图像分析与理解任务中,人们通常只对图像中的一小部分内容感兴趣,因此,人类视觉选择性注意机制的引入,不仅能将有限的计算资源集中到图像中的感兴趣区域,保证图像处理的效率,而且可以有效抑制图像中的无关信息,大大增强图像分析与理解系统的处理效果。本文主要研究基于视觉注意机制的复杂场景中多个小目标的检测问题,通过深入分析和探讨图像中兴趣目标的特征,建立改进的目标检测模型,来提高兴趣目标在显著图中的显著程度,从而达到率先聚焦于兴趣目标,快速实现目标检测的目的。 本文首先分析视觉注意的相关研究现状,介绍视觉注意机制的特点及其建模的理论基础和计算过程,对当前广泛应用的自底向上的Itti视觉注意计算模型进行重点研究。显著图是视觉注意的基础,有意识地增强兴趣目标在显著图中的响应是快速实现目标检测的一种途径。针对Itti模型在复杂背景的干扰下不能注意到所有小目标的不足,本文对Itti模型进行了改进,并通过实验验证改进后的模型能大大增强兴趣目标在显著图中的显著度。针对图像中部分兴趣目标与背景的对比度较弱的问题,本文通过对预处理后的图像在多尺度上建立侧抑制网络模型,生成对比度特征映射图,进而融合两幅显著性图得到总显著图。     显著图生成后,注意焦点在各兴趣目标间进行转移。为了在图像中完整地标出各兴趣目标,使目标具有最佳表达尺度,本文对图像的预注意区域使用模糊C-均值聚类算法进行分割。通过仿真实验表明,各兴趣目标能被完整地标记出来,检测结果符合人类的实际视觉认知。     本文还对视觉注意计算模型的FPGA实现做了初步尝试,提出了相应的流程框图,对其中重要模块的实现原理和过程进行详细的介绍,并给出仿真结果。
英文摘要: With the rapid development of information technology, image is playing an increasing role as one kind of information carrier. It provides a huge amount of information for people, but also brings difficulties for manual processing. In a wide variety of tasks for image analysis and understanding, people usually are interested in a small part of the contents of the image. Therefore, to bring in the human visual selective attention mechanism can not only focus on the region of interest with the limited computing resources, which ensures the efficiency of image processing, but also inhibit the irrelevant information in the image effectively, which enhances the treatment effect of the image analysis and understanding system greatly. This paper mainly studies the detection of small multi-target in complex scene based on visual attention mechanism. We improve the degree of saliency for the interested targets in the saliency map by deeply analysing and exploring the feature of the interested targets in the image and then establishing modified target detection model. The model can help us focus on the interested targets firstly to achieve the goal of detecting targets quickly.     The paper analyzes the research status of visual attention firstly, and then introduces the characteristic of visual attention mechanism. Then the rationale and computational process of modeling about attention mechanism is introduced. The bottom-up Itti’s visual attention computation model is researched detailedly. Saliency map is the basis of visual attention, and enhancing the response of the interested targets in the saliency map intentionally is a way to detect targets rapidly. For the lack to Itti’s model of unable to notice all the small targets in cluster, Itti’s model has been improved in this paper, and it testifies that the improved model can enhance the degree of saliency of the interested targets greatly by experiments. For the contrast between part of the interested targets and the background is unconspicuous, a lateral inhibition network model is established on the multi-scale after the pre-process of the image. A contrast feature map is generated by the lateral inhibition network model, and then we get the final saliency map by combining the foregoing two saliency maps.     After the saliency map is generated, the focus of attention transfers between the interested targets. In order to mark the interested targets in the image integrallty and make the targets owe the best scale, fuzzy C-means clustering algorithm is applied to segment the pre-attention area in the image. The simulation result shows that the interested targets can be marked integrallty and it accords with the actual visual cognition of human. This paper also attempts to implement the visual attention computational model on FPGA, and the flow chart is proposed. The principle and process about some important modules are introduced in detail, and the simulation result is offered.
语种: 中文
产权排序: 1
内容类型: 学位论文
URI标识: http://ir.sia.cn/handle/173321/9239
Appears in Collections:光电信息技术研究室_学位论文

Files in This Item:
File Name/ File Size Content Type Version Access License
基于视觉注意的小目标检测方法研究.pdf(1503KB)----限制开放 联系获取全文

Recommended Citation:
谢文亮.基于视觉注意的小目标检测方法研究.[硕士 学位论文 ].中国科学院沈阳自动化研究所 .2012
Service
Recommend this item
Sava as my favorate item
Show this item's statistics
Export Endnote File
Google Scholar
Similar articles in Google Scholar
[谢文亮]'s Articles
CSDL cross search
Similar articles in CSDL Cross Search
[谢文亮]‘s Articles
Related Copyright Policies
Null
Social Bookmarking
Add to CiteULike Add to Connotea Add to Del.icio.us Add to Digg Add to Reddit
所有评论 (0)
暂无评论
 
评注功能仅针对注册用户开放,请您登录
您对该条目有什么异议,请填写以下表单,管理员会尽快联系您。
内 容:
Email:  *
单位:
验证码:   刷新
您在IR的使用过程中有什么好的想法或者建议可以反馈给我们。
标 题:
 *
内 容:
Email:  *
验证码:   刷新

Items in IR are protected by copyright, with all rights reserved, unless otherwise indicated.

 

 

Valid XHTML 1.0!
Copyright © 2007-2016  中国科学院沈阳自动化研究所 - Feedback
Powered by CSpace