A Study on the Methods for Estimating the Distribution of Pedestrians in an Underground Mall by Use of Watch Cameras

This paper shows the method of estimating spatiotemporal distribution of pedestrians by using watch cameras. We estimate the distribution without tracking technology, with pedestrian＇s privacy protected and in Umeda underground mall. Lately spatiotemporal distribution of pedestrians has being increasingly important in the field of urban planning, disaster prevention planning, marketing and so on. Although many researchers have tried to capture the information of location as dealing with some sensors, some problems still remain, such as the investment of sensors, the restriction of the number of people who has the device they are able to capture. From such background, we develop an original labelling algorithm and estimate the spatiotemporal distribution of pedestrians and the information of the passing time and the direction of pedestrians from sequential images of a watch camera.

机器人三维视觉实验平台坐标系标定实验设计

为了探究机器人工作站的高精度坐标系标定方法,搭建了一个由工业机器人、双目结构光相机以及单轴转台组成的机器人三维视觉实验平台。研究了定位误差补偿方法、工业机器人工具中心点标定方法、工业机器人与双目结构光相机的手眼标定方法以及双目结构光相机与单轴转台的坐标系标定方法,并形成了实验平台的坐标系综合标定方法,最后测试了标定后的综合精度。实验结果表明,工业机器人与双目结构光相机之间的平均标定误差为0.442~0.689 mm,机器人三维视觉实验平台的综合标定误差为0.399~0.696 mm,能完全满足焊接、涂胶、切割等机器人离线编程应用的精度要求。