Cooperative Distributed Image Understanding Systems

在本研究中，我们将协作式分布式处理(多智能体系统)作为实现健壮性和灵活性的图像理解系统的系统架构，并通过开发实用的协作式分布式图像理解系统来验证其有效性。本研究的主要结果如下：1.图像理解的分布式协同处理类型可分为以下几类：(A)分布式传感系统(B)分布式分析系统(b-1)函数分布(b-2)算法分布(b-3)协同空间推理。一种多智能体区域分割系统，其中每个图像区域被视为一个智能体3。用于获取全向图像的固定视点云台摄像机和将现成的活动摄像机建模为固定视点云台变焦摄像机的摄像机校准软件4。一种使用固定视点平移-倾斜-变焦摄像机的实时目标检测和跟踪系统。具有多个分布式固定视点云台摄像机的协同目标跟踪系统6。基于多个视频数据的协同分布式行为识别。一种由一组无公共视场的分布式观测站组成的广域目标跟踪系统

In this study we consider the cooperative distributed processing (multi-agent system) as a system architecture to realize robust and flexible image understanding systems, and demonstrate its effectiveness by developing practical cooperative distributed image understanding systems. The results of the study can be summarized as follow :1. The types of cooperative distributed processing for image understanding are classified into the following categories : (a) Distributed Sensing Systems (b) Distributed Analysis Systems ((b-1) Functional Distribution (b-2) Algorithm Distribution (b-3) Cooperative Spatial Reasoning).2. A multi-agent region segmentation system, where each image region is regarded as an agent3. A fixed viewpoint pan-tilt camera to acquire ominidirectional images and a camera calibration software to model an off-the-shelf active camera as a fixed viewpoint pan-tilt-zoom camera4. A real time object detection and tracking system using a fixed viewpoint pan-tilt-zoom camera5. A cooperative object tracking system with multiple distributed fixed viewpoint pan-tilt-zoom cameras6. A cooperative distributed behavior recognition from multiple video data7. A wide area object tracking system consisting a group of distributed observation stations without common visual fields