{"created":"2021-03-01T06:22:07.950406+00:00","id":5436,"links":{},"metadata":{"_buckets":{"deposit":"701a7369-409c-4c24-a81d-8397c0d438e9"},"_deposit":{"id":"5436","owners":[],"pid":{"revision_id":0,"type":"depid","value":"5436"},"status":"published"},"_oai":{"id":"oai:repository.dl.itc.u-tokyo.ac.jp:00005436","sets":["6:209:392","9:233:280"]},"author_link":["11350"],"control_number":"5436","item_7_alternative_title_1":{"attribute_name":"その他のタイトル","attribute_value_mlt":[{"subitem_alternative_title":"オンライン形状学習機能を備えた実時間物体追跡システム"}]},"item_7_biblio_info_7":{"attribute_name":"書誌情報","attribute_value_mlt":[{"bibliographicIssueDates":{"bibliographicIssueDate":"2013-03-25","bibliographicIssueDateType":"Issued"}}]},"item_7_date_granted_25":{"attribute_name":"学位授与年月日","attribute_value_mlt":[{"subitem_dategranted":"2013-03-25"}]},"item_7_degree_grantor_23":{"attribute_name":"学位授与機関","attribute_value_mlt":[{"subitem_degreegrantor":[{"subitem_degreegrantor_language":"ja","subitem_degreegrantor_name":"東京大学"},{"subitem_degreegrantor_language":"en","subitem_degreegrantor_name":"The University of Tokyo"}],"subitem_degreegrantor_identifier":[{"subitem_degreegrantor_identifier_name":"12601","subitem_degreegrantor_identifier_scheme":"kakenhi"}]}]},"item_7_degree_name_20":{"attribute_name":"学位名","attribute_value_mlt":[{"subitem_degreename":"博士(工学)"}]},"item_7_description_5":{"attribute_name":"抄録","attribute_value_mlt":[{"subitem_description":"Object tracking plays an important role in many applications, such as video surveillance, human-computer interaction, vehicle navigation, and robot control. It is generally defined as a task of estimating the location of an object over a sequence of images. In practical applications, there are many factors that make the task complex such as illumination variation, appearance change, shape deformation, partial occlusion, and camera motion. Moreover, lots of these applications require real-time response. Therefore, the development of real-time working algorithms is of essential importance. In order to accomplish such a challenging task, a real-time tracking system has been developed and proposed in this thesis. In this thesis, a solution to the tracking task is proposed based on consideration of efficient implementation as priority, and several critical issues are resolved as follows. At first, a hardware-friendly tracking framework is designed, which is implemented on field-programmable gate array (FPGA) technology, and compatible with very large scale integration (VLSI) technology. This framework, named multiple candidate regeneration (MCR), is developed as a simple but high-speed and high-efficiency searching algorithm. The basic idea was inherited from the particle filter (PF) but the algorithm has been greatly modified from the original particle filter so that it can be implemented on VLSI hardware very efficiently. The important difference between MCR and PF is that the MCR is developed by simplifying the visual tracking task and considering the simple hardware implementation. It can be considered as an efficient searching strategy instead of an intensity estimation method. In the development, several problems that may limit the hardware performance have been solved, such as complex computation, data transmission and utilization of hardware resources. The proposed architecture achieved 150 frame per second (f/s) on FPGA, and can reach about 900 f/s if it is implemented on VLSI with on-chip image sensor. This solution has several advantages. First, it works at high frame rate, which can enhance the effect of localization. It also meets the requirement of higher processing speed in some complex intelligence systems, which seems difficult to achieve by conventional solutions. Second, the system can be extended to be useful in many applications because of its flexibility. Third, since the processing speed is faster than the frame rate, there is still large space for further improving the ability of the system without losing real-time performance. The system was implemented on a Terasic DE3 FPGA board. Under the operating frequency of 60 MHz, the experimental system achieved a processing ability of 0.8 ms per frame in tracking a 64 * 64 scale object image in 640 * 480-pixel size video sequences. In tracking algorithms, how to represent the target image is of particular importance because it greatly influences the tracking performance under certain tracking framework. Color, edge, and texture are typical attributes used for representing objects. A number of other features, including active contour, scale-invariant feature transform (SIFT) feature, oriented energy, and optical flow, are also used in many works. Some works also combine these features or incorporate on-line learning of the model of an object and background. In this thesis, we have aimed to establish both robustness of object representation and the real-time performance of the processing, because feature extraction is usually a time consuming process. It was well known that the visual perception of animals relies heavily on the directional edges. In the present work, therefore, the directional-edge-based image feature representation algorithm is employed to represent the object image. Robust performance of the directional-edge-based algorithms has already been demonstrated in various image recognition applications. In addition, dedicated VLSI chips for efficient directional edge detection and image vector generation have also been developed for object recognition systems. Whether a tracking system can be easily extended for various purposes is also a critical issue. This thesis contains a detailed discussion on extending the function of the system, including hardware implementation on VLSI, multiple-object tracking, full-occlusion and initialization problems, and employing of state vector. The architecture of this system is compatible with VLSI design, and may reach better performance on VLSI. For the multiple-object tracking, an efficient method is proposed to allocate the limited hardware resources. For the fullocclusion and initialization problems, a searching algorithm based on proposed system is developed. By using the state vector, more attributes can be estimated for achieving more information about the object, which also helps deal with appearance change and increase the tracking accuracy. The following parts of the thesis are focused on building learning ability for the system. For object tracking, one promising direction is to consider the object tracking as a binary classification problem, and employ discriminative methods in the tracking framework. Nearest neighbor (NN) classifier is a simple but widely used classifier. Some tracking algorithms have tried to use it because of its effectiveness in some tasks and its outstanding simplicity. Support vector machine (SVM), as a powerful classification scheme, has been also used in many tracking algorithms, benefiting the algorithms with accurate localization and flexible modeling of the target. Each of the classifier works as an appearance model of the target by changing its templates while training. For NN classifier, the training and testing process is really simple and fit for hardware implementation. For the SVM classifier, one feature is that the boundary is represented by the combination of support vectors, and the number of support vectors is usually a small portion of the total training dataset. This feature becomes very important when implementing the tracking algorithm on hardware, because the hardware resources are always limited. The SVM-based tracking system proposed in this thesis aims to solve the following problems. Some work builds a superior SVM classifier and gives good results in tracking vehicles. However, the off-line training mechanism employed in the work requires a large number of training samples selected manually and does not support updating the training samples. In some research, all samples learned from each frame of an image sequence are stored for training the SVM. This causes a large memory cost if it is used in a long-duration task. In some work, a simple strategy is employed to determine new training samples, which may cause “drift problem”. Moreover, these algorithms do not consider their real-time performances, which is in fact of great importance in object tracking applications. This is mainly because of the complex computation of SVM. Especially for the on-line learning SVMs, frequently repeated training and predictions make this problem even worse. Therefore, in order to extend the power of SVM in most of the general tracking applications, it is necessary to develop a proper tracking framework and a VLSI hardware-implementation friendly structure for the SVM-based algorithm. A real-time visual tracking algorithm is presented employing an online support vector machine scheme. In this system, a novel training framework is proposed, which enables the system to select reliable training samples from the image sequence. The tracking framework includes how to update training samples and how to select test samples and make prediction of the target location. Different from other algorithms, this framework gives a rule guiding the selection of target training samples. When the target changes its appearance significantly, the system may fail to localize the target because the classifier misclassifies the target image to the background image category. In order to solve this problem, background samples are utilized to predict the location of the target image. Unlike the moving target image, most of the background sample images are stable. As a result, highaccuracy tracking has been established. In addition, regarding the selection of target samples for on-line training of SVM, a new selection method has been introduced. The on-line SVM learning requires repeated training and predicting process. The predicting process always contains computation of thousands of test samples in conventional algorithms, preventing these algorithms from working in real-time. In this process, not only the SVM, but also the feature extraction of each sample will cost lots of time. Based on a SVM chip developed in our group, the most complex part in this algorithm can be computed efficiently. At the same time, multiple candidate regeneration is employed to reduce the computational cost without sacrificing the tracking accuracy. In addition, the directional-edge-feature vector representation, whose VLSI implementation has been proposed, is employed to represent the sample images. The algorithm has been evaluated on challenging video sequences and showed robust tracking ability with accurate tracking results. The hardware implementation is also discussed, while verification has been done to prove the real-time ability of this algorithm. After development of the SVM-based tracking system, the essential facts of the tracking task were analyzed, and an NN-based tracking system has been proposed. The basic idea is that a classifier specially designed for tracking task is more efficient. The classifiers mentioned above are usually used in object recognition in computer vision. Compared with object recognition task, object tracking contains much less categories of objects, and it is obvious that object recognition is a time-consuming work in most of the time. Therefore, a “weak classifier to recognition can be sufficient for the tracking task, and it is also very important to the hardware implementation. In this part, relationship between similarity, APED vector, and NN classifier is analyzed. Based on the analysis, a new appearance model use basic NN has been proposed. The accuracy of this system has been evaluated and the simplicity of hardware implementation has been discussed. In summary, this thesis presents a novel real-time solution to object tracking task with learning ability. The robust feature learning ability of the system is realized by introducing the SVM and NN classifiers into the tracking system, and designing a new tracking framework for the classifier-based algorithm. The hardware implementation problem was considered carefully. Hardware-friendly architecture has been designed and the real-time tracking system has been finally implemented on an FPGA board with a dedicated VLSI chip. Extensive experiments have been performed for evaluation on the tracking accuracy of this system. The thesis also contains very detailed discussions about the system.","subitem_description_type":"Abstract"}]},"item_7_identifier_registration":{"attribute_name":"ID登録","attribute_value_mlt":[{"subitem_identifier_reg_text":"10.15083/00005427","subitem_identifier_reg_type":"JaLC"}]},"item_7_select_21":{"attribute_name":"学位","attribute_value_mlt":[{"subitem_select_item":"doctoral"}]},"item_7_text_22":{"attribute_name":"学位分野","attribute_value_mlt":[{"subitem_text_value":"Engineering (工学)"}]},"item_7_text_24":{"attribute_name":"研究科・専攻","attribute_value_mlt":[{"subitem_text_value":"Department of Electrical Engineering and Information Systems, Graduate School of Engineering (工学系研究科電気系工学専攻)"}]},"item_7_text_4":{"attribute_name":"著者所属","attribute_value_mlt":[{"subitem_text_value":"東京大学大学院工学系研究科電気系工学専攻"},{"subitem_text_value":"Department of Electrical Engineering and Information Systems, Graduate School of Engineering, The University of Tokyo"}]},"item_creator":{"attribute_name":"著者","attribute_type":"creator","attribute_value_mlt":[{"creatorNames":[{"creatorName":"趙, 普社"}],"nameIdentifiers":[{"nameIdentifier":"11350","nameIdentifierScheme":"WEKO"}]}]},"item_files":{"attribute_name":"ファイル情報","attribute_type":"file","attribute_value_mlt":[{"accessrole":"open_date","date":[{"dateType":"Available","dateValue":"2017-06-01"}],"displaytype":"detail","filename":"37097414.pdf","filesize":[{"value":"31.8 MB"}],"format":"application/pdf","licensetype":"license_note","mimetype":"application/pdf","url":{"label":"37097414.pdf","url":"https://repository.dl.itc.u-tokyo.ac.jp/record/5436/files/37097414.pdf"},"version_id":"8c087a74-9d23-4d49-9969-22ed051546a2"}]},"item_language":{"attribute_name":"言語","attribute_value_mlt":[{"subitem_language":"eng"}]},"item_resource_type":{"attribute_name":"資源タイプ","attribute_value_mlt":[{"resourcetype":"thesis","resourceuri":"http://purl.org/coar/resource_type/c_46ec"}]},"item_title":"A Real-Time Object Tracking System With On-Line Feature Learning","item_titles":{"attribute_name":"タイトル","attribute_value_mlt":[{"subitem_title":"A Real-Time Object Tracking System With On-Line Feature Learning","subitem_title_language":"en"}]},"item_type_id":"7","owner":"1","path":["280","392"],"pubdate":{"attribute_name":"PubDate","attribute_value":"2013-10-02"},"publish_date":"2013-10-02","publish_status":"0","recid":"5436","relation_version_is_last":true,"title":["A Real-Time Object Tracking System With On-Line Feature Learning"],"weko_creator_id":"1","weko_shared_id":-1},"updated":"2023-03-10T01:57:27.472234+00:00"}