SIA OpenIR  > 机器人学研究室
线状目标识别算法研究及应用
Alternative TitleResearch and Application of Linear Object Recognition Algorithm
赵利坡1,2
Department机器人学研究室
Thesis Advisor唐延东
ClassificationTP391.4
Keyword线状目标 多尺度线状目标强化 Radon变换 主动轮廓 开曲线检测
Call NumberTP391.4/Z45/2011
Pages57页
Degree Discipline模式识别与智能系统
Degree Name硕士
2011-05-27
Degree Grantor中国科学院沈阳自动化研究所
Place of Conferral沈阳
Abstract线状目标识别在计算机视觉和图像理解领域起着重要的作用,对线状目标识别进行研究不仅具有深远的理论意义,也具有广泛的实用价值。线状目标识别是把图像分成各具线状特性的区域并提取感兴趣的线状目标的技术和过程,是图像处理与计算机视觉中的重要组成部分,也是理解图像的重要途径,更是图像跟踪、匹配等研究领域及应用的一个必要环节。 本论文主要是对线状目标识别算法进行研究探讨,并结合两个实际应用项目进行研究。在对国内外相关研究领域学习、分析的基础上,针对应用中待识别目标的不同特性设计了不同的线状目标识别算法。本论文主要工作内容及成果有如下四部分: 1、介绍线状目标识别中常用的经典算法,探讨他们各自的性能以及适用范围,对多尺度线状目标强化算法、Hough变换、Radon变换、主动轮廓模型、CV模型算法提取线状目标进行了讨论,从算法精度和实时性两方面分析了他们的性能。 2、以巡线无人机巡航中识别高压输电线为背景,构建基于DSP的输电线路自动巡检视觉系统平台,包括机载视觉子系统、地面站子系统、无线通信子系统,构建了完整的空地、人机交互环路等。 3、针对高压输电线路的成像特征,提出了一种准确、实时的高压输电线检测与识别算法,从准确性和实时性两方面分别提出了基于方向约束的多尺度线状目标强化算子和基于角度约束的Radon变换算子等。 4、以测量路面平整度为背景,构建基于线激光的平整度自动检测系统平台。针对激光线成像特性,提出基于亮度的主动轮廓模型开曲线检测方法。本算法引入了灰度约束,使曲线向图像最亮的地方演化。本算法的优点是演化的轮廓是开曲线,而一般主动轮廓模型演化的是整个闭合曲线,而激光线也正是贯穿整幅图像的开曲线。为了保证检测的准确性,对抑制物体边缘干扰进行了研究,提出了边缘抑制算子,从而保证检测激光线的准确性。经过大量的机场跑道和公路路面图像实验证明,本算法不仅能准确识别出路面图像中的激光线,而且还具有很好的抗干扰性。
Other AbstractLinear object recognition plays an important role in the field of computer vision and image understanding. The study of linear object recognition has not only profound theoretical significance, also extensive practical value. Linear object recognition is the technology and process that divide an image into some areas with different linear characteristics and extract interested linear objects. It is an important part in image processing and computer vision. It is also the important way to understand the image, but also a necessary link in the fields such as image tracking and image matching. This paper is mainly to the linear object recognition algorithm to study, combined with two practical projects. According to the different characteristics of identifying objects, we design different linear object recognition algorithms based on the study and analysis in the domestic and foreign relevant fields. The main contents and contributions are summarized as follows. 1. We introduce classical linear object recognition algorithms and investigate their performance and application scope in this paper. In order to extract linear objects, some algorithms are discussed such as Multi-scale linear object enhancement, Hough transform, Radon transform, Active contour model and CV model in precision and real-time. 2. We design a visual system platform based on DSP for the inspection of high-voltage transmission line with Unmanned Aerial Vehicle (UAV) .The system includes airborne vision subsystem, sites subsystem and wireless communications subsystem constructing a complete air to ground and human-computer interaction circuit. 3. According to the imaging characteristics of high-voltage transmission line, we present an accurate, real-time high-voltage transmission line detection and recognition algorithm for the inspection of high-voltage transmission line. Meanwhile, multi-scale linear object enhancement with direction constraint and radon transform based on the angle constraint is proposed respectively from the aspects of accuracy and real-time in this paper. 4. We design a system platform for automatic detecting the pavement roughness with the line laser. According to the imaging characteristics of the laser line projected onto the road, we propose a new open curve detection algorithm via active contour method based on image brightness. The laser line has the characteristics that its gray value is greater than that of background, so the proposed algorithm has a gray value constraint to drive the evolution curve to the brightest place in image. The laser line on the road is the open curve throughout the whole image, so evolution outline of our algorithm is open curve while general active contour model is closed curve. Moreover, we design an edge inhibit operator to restrain the influence of objects edges to ensure the accuracy of the test laser line. Numerical experiments utilizing the airport runway and highway image present our algorithm not only can detect the laser line correctly but also is robust to noise immunity.
Language中文
Contribution Rank1
Document Type学位论文
Identifierhttp://ir.sia.cn/handle/173321/9408
Collection机器人学研究室
Affiliation1.中国科学院沈阳自动化研究所
2.中国科学院研究生院
Recommended Citation
GB/T 7714
赵利坡. 线状目标识别算法研究及应用[D]. 沈阳. 中国科学院沈阳自动化研究所,2011.
Files in This Item:
File Name/Size DocType Version Access License
线状目标识别算法研究及应用.pdf(4738KB) 开放获取LicenseApplication Full Text
Related Services
Recommend this item
Bookmark
Usage statistics
Export to Endnote
Google Scholar
Similar articles in Google Scholar
[赵利坡]'s Articles
Baidu academic
Similar articles in Baidu academic
[赵利坡]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[赵利坡]'s Articles
Terms of Use
No data!
Social Bookmark/Share
All comments (0)
No comment.
 

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