双目变焦调焦系统
摘要
双目视觉系统通常采用的是定焦距的摄像机。但在很多应用中,需要在不改变双目视觉系统位置的情况下,既能清晰的观察到远距离对象,也可以观察近距离的对象,定焦距的双目视觉系统很难适应这一需求。而变焦距的双目视觉系统可以解决这一问题。变焦距的双目视觉系统具有定焦距系统不具有的灵活性和方便性,可以广泛应用于立体摄影、运动跟踪、移动机器人、三维检测、人机交互等。双目变焦系统的难点在于如何确保两个摄像机焦距的同步,以及如何在摄像机的焦距改变之后,更新摄像机的参数。当然,同单目变焦系统一样,自动聚焦也是双目变焦系统的关键问题。
首先,本文研究了单目系统中自动聚焦的基本原理,分析了自动聚焦的评价函数、搜索策略和聚焦窗口的选择,并改进了自动聚焦评价函数,使之获得更好的聚焦效果。改进的自动聚焦评价函数采用绝对方差函数抑制能量梯度函数的噪声和多峰性,得到了高灵敏度的单峰的自动聚焦评价函数。
其次,本文研究了双目视觉系统中双摄像机焦距同步的问题。针对双摄像机系统对硬件要求高、安装设计复杂的问题提出了基于图像匹配的双摄像机焦距同步方法,使得变焦距的双目视觉系统具有更好的泛用性和灵活性。
最后,本文设计并实现了双目变焦调焦系统,可以进行自动聚焦,并保持焦距同步。其中对系统的硬件及软件设计进行了详细的描述,同时也针对自动聚焦和焦距同步算法在系统实现的过程中出现的问题作了具体分析。最终系统可以有效的进行自适应调节,保持两个摄像机聚焦清晰以及焦距同步。
Binocular Vision System is commonly basedon fixedfocus camera. But in many cases, We need to observe both the distant objects and the objects at close range clearly without changing the location of Binocular Vision System. It is difficult for the Fixed Focus Binocular Vision System to meet this need. And the Variable Focus Binocular Visual System can solve this problem. Auto Zoom and Focus Binocular Vision System are flexible and convenient which the fixed focus system is not. It can be widely used in three-dimensional photography, motion tracking, mobile robotics, three-dimensional detection, human-computer interaction. One problem of the Binocular Zoom System is how to keep the focus length of the two cameras synchronous. Another is how to update camera parameters after the lensfocus changed. Certainly, as the same system with monocular zoom, auto focus is a key issue to the Binocular Zoom System.
First, the basic theories and methods of auto focus in the Single Camera System are studied. And the focus measure function, the focus place search algorithm and intelligent focus framing technologie are analysised. In order to obtain a better result offocusing, the focus measure function is improved. The improved focus measure function use the Absolute variance function to suppress the noise and multi-modalnature of the energy gradient function, and it is with highly sensitive and single-peak.
Secondly, the zoom synchronization of Binocular Zoom System is studied. To solve the problem which Variable Focus Binocular Visual System sets high requirements for hardware and is very complex in installation and design. This paper give a method of zoom synchronization based on the image matching which makes Variable Focus Binocular Visual System more flexible and with higher Generality.
Finally, Auto Zoom and Focus Binocular Vision System is designed, which can focus automatically and keep the focal length synchronized. In this paper, the design of the hardware and the software about this system is described in detail.And the problem appeared in the design of the system about the auto focus and zoom synchronization is analysised concretely. The system can be adaptive to adjust itself effectively to keep the two cameras focused and the focal length synchronized.
首先,本文研究了单目系统中自动聚焦的基本原理,分析了自动聚焦的评价函数、搜索策略和聚焦窗口的选择,并改进了自动聚焦评价函数,使之获得更好的聚焦效果。改进的自动聚焦评价函数采用绝对方差函数抑制能量梯度函数的噪声和多峰性,得到了高灵敏度的单峰的自动聚焦评价函数。
其次,本文研究了双目视觉系统中双摄像机焦距同步的问题。针对双摄像机系统对硬件要求高、安装设计复杂的问题提出了基于图像匹配的双摄像机焦距同步方法,使得变焦距的双目视觉系统具有更好的泛用性和灵活性。
最后,本文设计并实现了双目变焦调焦系统,可以进行自动聚焦,并保持焦距同步。其中对系统的硬件及软件设计进行了详细的描述,同时也针对自动聚焦和焦距同步算法在系统实现的过程中出现的问题作了具体分析。最终系统可以有效的进行自适应调节,保持两个摄像机聚焦清晰以及焦距同步。
Binocular Vision System is commonly basedon fixedfocus camera. But in many cases, We need to observe both the distant objects and the objects at close range clearly without changing the location of Binocular Vision System. It is difficult for the Fixed Focus Binocular Vision System to meet this need. And the Variable Focus Binocular Visual System can solve this problem. Auto Zoom and Focus Binocular Vision System are flexible and convenient which the fixed focus system is not. It can be widely used in three-dimensional photography, motion tracking, mobile robotics, three-dimensional detection, human-computer interaction. One problem of the Binocular Zoom System is how to keep the focus length of the two cameras synchronous. Another is how to update camera parameters after the lensfocus changed. Certainly, as the same system with monocular zoom, auto focus is a key issue to the Binocular Zoom System.
First, the basic theories and methods of auto focus in the Single Camera System are studied. And the focus measure function, the focus place search algorithm and intelligent focus framing technologie are analysised. In order to obtain a better result offocusing, the focus measure function is improved. The improved focus measure function use the Absolute variance function to suppress the noise and multi-modalnature of the energy gradient function, and it is with highly sensitive and single-peak.
Secondly, the zoom synchronization of Binocular Zoom System is studied. To solve the problem which Variable Focus Binocular Visual System sets high requirements for hardware and is very complex in installation and design. This paper give a method of zoom synchronization based on the image matching which makes Variable Focus Binocular Visual System more flexible and with higher Generality.
Finally, Auto Zoom and Focus Binocular Vision System is designed, which can focus automatically and keep the focal length synchronized. In this paper, the design of the hardware and the software about this system is described in detail.And the problem appeared in the design of the system about the auto focus and zoom synchronization is analysised concretely. The system can be adaptive to adjust itself effectively to keep the two cameras focused and the focal length synchronized.
引文
[1]贾云得.机器视觉,科学出版社, 1999.
[2]张乐.自动聚焦系统精确性和实时性的研究[D].山东大学硕士学位论文, 2008
[3]李奇.数字自动对焦技术的理论及实现方法研究[D].浙江大学博士学位论文, 2004
[4]康宗明,张利,谢攀.一种基于能量和熵的自动聚焦算法[J].电子学报, 2003, 31(4):23-25
[5]庞胜利.图像模糊度评价研究[D].西安电子科技大学硕士学位论文, 2010
[6]陶青川,邓宏彬.基于小波变换的高斯点扩展函数估计[J].光学技术, 2004, 30(3):284-288
[7]徐杰,陈一民,史志龙.双目视觉变焦测距技术[J].上海大学学报(自然科学版), 2009, 15(2):160-174
[8]徐杰.基于双目视觉的运动机器人自动追踪与避障技术的研究[D].上海大学硕士学位论文, 2008
[9]余蓉.自动变焦聚焦视觉传感器及其标定问题研究[D].武汉理工大学硕士学位论文, 2007
[10]陈琛.图像式三坐标测量仪大范围快速自动调焦策略的研究[D].上海交通大学硕士学位论文, 2008
[11]蒋婷.谭跃刚.刘泉基于SOBEL算子的图像清晰度评价函数研究[J].计算机与数字工程2008(8):
[12]陈国金.朱妙芬.张克松图像调焦过程的清晰度评价函数研究[J].数据采集与处理2009(2)
[13]陈国金.数字图像自动聚焦技术研究及系统实现[D].西安电子科技大学博士学位论文, 2007
[14]李岩琪.基于SIFT算子的双目视觉立体匹配算法研究[D].西安电子科技大学硕士学位论文, 2010
[15]吴铮.基于特征的图像配准算法研究[D].浙江大学硕士学位论文,2006
[16]张耀宗.图像匹配算法的研究与应用[D].太原理工大学硕士学位论文,2010
[17]徐秀云.基于特征点的景象匹配技术研究[D].南京理工大学硕士学位论文,2009
[18]谢东海,詹总谦,江万寿.改进的harris算子用于点特征的精确定位[D].南京理工大学硕士学位论文,2009
[19]杨占龙.基于特征点的图像配准与拼接技术研究[D].西安电子科技大学博士学位论文, 2008
[20]李鸣翔,贾云得.基于自适应聚合的立体视觉合作算法[J].软件学报,2008,19(7): 1674-1682
[21]MARR D, Yao Guozheng translation.Vision Computing Theory.Beijing: Higher Education Press, 1988
[22]Chia-Hao Chang, Chiou-Shann Fuh. Auto Focus Using Adaptive Step Size Search and Zoom Tracking Algorithm[J].ICASSP,2006
[23]Ni Jun, Auto-focus Algorithm for Moving Target Image[J].ICIG,2009:728-732
[24]Yong-Sheng Chen, Yi-Ping Hung,Chiou-ShannFuh. A Fast Block Matching Algorithm Based on the Winner-Update Strategy[J].Image Processing2002:1212-1222
[25]Kazushige Ooi, Keiji Izumi, Mitsuyuki Nozaki, Isao Takeda, An advanced autofocus system for video camera using quasi condition reasoning[J].Consumer Electronics, 1990:526-530
[26]Dingrui Wan, Jie Zhou, Multiresolution and Wide-Scope Depth Estimation Using a Dual-PTZ-Camera System[J].Image Processing,2009: 677 - 682
[27]N. K. Chern, N. P. A. Neow, and M. H. Ang,Jr.,Practical Issues in Pixel-Based Autofocusing for Machine Vision[J].Robotics and Automation,2001: 2791 - 2796 vol.3
[28]Li Cheng Chiu,ChiouShannFuh.An Efficient Auto Focus Method for Digital Still CameraBased on Focus Value Curve Prediction Model[J].Consumer Electronics,2010: 1135-1143
[29]JufieF,SchmidC.Scale-invariant shape features for recognition of object categories [J].CVPR,2004,(2):90—96
[30]Jaechul Kim,Grauman K.Asymmetric region-to-image matching for comparing images with generic object categories[J].CVPR,2010:2344-2351
[31]Yang Li-mei.Rotation resistance airborne remote image real time matching algorithm[J].CMCE, 2010:502-505
[32]YupingFeng, Shi Li, Ming Dai. An Image Matching Algorithm Based on Sub-block Coding[J].WCSE,2009:599 - 603
[33]ZhangTianxu,LuWeixue,PengJiaxiong.Adaptive image matching via spatial varying gray-level correction[J].IEEE Trans on Communication,1995:1970-1980
[34]Chiou-ShannFuh,Homg-bin Liu.Projection for patem recognition.Image and VisionComputing.1998:677-687
[2]张乐.自动聚焦系统精确性和实时性的研究[D].山东大学硕士学位论文, 2008
[3]李奇.数字自动对焦技术的理论及实现方法研究[D].浙江大学博士学位论文, 2004
[4]康宗明,张利,谢攀.一种基于能量和熵的自动聚焦算法[J].电子学报, 2003, 31(4):23-25
[5]庞胜利.图像模糊度评价研究[D].西安电子科技大学硕士学位论文, 2010
[6]陶青川,邓宏彬.基于小波变换的高斯点扩展函数估计[J].光学技术, 2004, 30(3):284-288
[7]徐杰,陈一民,史志龙.双目视觉变焦测距技术[J].上海大学学报(自然科学版), 2009, 15(2):160-174
[8]徐杰.基于双目视觉的运动机器人自动追踪与避障技术的研究[D].上海大学硕士学位论文, 2008
[9]余蓉.自动变焦聚焦视觉传感器及其标定问题研究[D].武汉理工大学硕士学位论文, 2007
[10]陈琛.图像式三坐标测量仪大范围快速自动调焦策略的研究[D].上海交通大学硕士学位论文, 2008
[11]蒋婷.谭跃刚.刘泉基于SOBEL算子的图像清晰度评价函数研究[J].计算机与数字工程2008(8):
[12]陈国金.朱妙芬.张克松图像调焦过程的清晰度评价函数研究[J].数据采集与处理2009(2)
[13]陈国金.数字图像自动聚焦技术研究及系统实现[D].西安电子科技大学博士学位论文, 2007
[14]李岩琪.基于SIFT算子的双目视觉立体匹配算法研究[D].西安电子科技大学硕士学位论文, 2010
[15]吴铮.基于特征的图像配准算法研究[D].浙江大学硕士学位论文,2006
[16]张耀宗.图像匹配算法的研究与应用[D].太原理工大学硕士学位论文,2010
[17]徐秀云.基于特征点的景象匹配技术研究[D].南京理工大学硕士学位论文,2009
[18]谢东海,詹总谦,江万寿.改进的harris算子用于点特征的精确定位[D].南京理工大学硕士学位论文,2009
[19]杨占龙.基于特征点的图像配准与拼接技术研究[D].西安电子科技大学博士学位论文, 2008
[20]李鸣翔,贾云得.基于自适应聚合的立体视觉合作算法[J].软件学报,2008,19(7): 1674-1682
[21]MARR D, Yao Guozheng translation.Vision Computing Theory.Beijing: Higher Education Press, 1988
[22]Chia-Hao Chang, Chiou-Shann Fuh. Auto Focus Using Adaptive Step Size Search and Zoom Tracking Algorithm[J].ICASSP,2006
[23]Ni Jun, Auto-focus Algorithm for Moving Target Image[J].ICIG,2009:728-732
[24]Yong-Sheng Chen, Yi-Ping Hung,Chiou-ShannFuh. A Fast Block Matching Algorithm Based on the Winner-Update Strategy[J].Image Processing2002:1212-1222
[25]Kazushige Ooi, Keiji Izumi, Mitsuyuki Nozaki, Isao Takeda, An advanced autofocus system for video camera using quasi condition reasoning[J].Consumer Electronics, 1990:526-530
[26]Dingrui Wan, Jie Zhou, Multiresolution and Wide-Scope Depth Estimation Using a Dual-PTZ-Camera System[J].Image Processing,2009: 677 - 682
[27]N. K. Chern, N. P. A. Neow, and M. H. Ang,Jr.,Practical Issues in Pixel-Based Autofocusing for Machine Vision[J].Robotics and Automation,2001: 2791 - 2796 vol.3
[28]Li Cheng Chiu,ChiouShannFuh.An Efficient Auto Focus Method for Digital Still CameraBased on Focus Value Curve Prediction Model[J].Consumer Electronics,2010: 1135-1143
[29]JufieF,SchmidC.Scale-invariant shape features for recognition of object categories [J].CVPR,2004,(2):90—96
[30]Jaechul Kim,Grauman K.Asymmetric region-to-image matching for comparing images with generic object categories[J].CVPR,2010:2344-2351
[31]Yang Li-mei.Rotation resistance airborne remote image real time matching algorithm[J].CMCE, 2010:502-505
[32]YupingFeng, Shi Li, Ming Dai. An Image Matching Algorithm Based on Sub-block Coding[J].WCSE,2009:599 - 603
[33]ZhangTianxu,LuWeixue,PengJiaxiong.Adaptive image matching via spatial varying gray-level correction[J].IEEE Trans on Communication,1995:1970-1980
[34]Chiou-ShannFuh,Homg-bin Liu.Projection for patem recognition.Image and VisionComputing.1998:677-687