SBIR Phase I: A Multi-Media Data Structure for 3D Measurement of Human/Animal Motion

这个小型企业创新研究(SBIR)第一阶段项目将探索开发多媒体数据结构的可行性，以模拟和存储活的、运动的生物的几何测量和运动特征。目标是开发用于从数字视频流中存储和访问关于移动动物的数据的工具，以便提取测量数据，并允许三维重建对象以及其在时间上的位置或作为动作位置(例如步幅位置)的函数，以及在步幅空间内导航信息树的高效算法。用卷尺测量任何动物，包括人类，都会有相当大的误差。拟议的项目旨在研究一项大胆而雄心勃勃的计划的可行性，如果成功，将解决这一问题和许多同类问题，并开辟一个具有相当大研究和发展潜力的领域。第一阶段的研究将集中于证明在正常行走速度下从动物(特别是马)那里获得一组测量数据的想法的可行性，以及足够的导航工具，以允许用户筛选3D视频流，以进行所需的调整，以获得完全跟踪的模型。目前对人类和动物进行静态研究和测量的方法要么是低技术含量的磁带，要么是需要大隔间的全身激光扫描仪或主动光扫描仪。即使在获得全身扫描后，就像服装业一样，由于缺乏方便的地标，获得所需的测量用于应用也是困难的。地标通过运动变得可见。目标是利用运动及其周期性特征来识别地标并恢复，不仅是主体的一个单一位置的‘雕像’，而且是它的整个运动。在美国，大约有1,000万头奶牛，其中约34%，即340万头奶牛每年是通过卷尺或猜测来衡量的。快速而准确地获得测量结果的能力将在乳制品行业以及纯种畜牧业中随时可用。

This Small-Business Innovation Research (SBIR) Phase I project will explore the feasibility of developing a multi-media data structure to model and store geometric measurement and motion characteristics of a living, moving creature. The objective is to develop tools for storing and accessing data on moving animals from digital video streams in order to extract measurement data and allow the three-dimensional reconstruction of the subject as well as its position in time or as a function of action position (such as stride position), together with efficient algorithms to navigate the information tree within the stride space. Measurement by tape of any animal, including humans, is subject to sizable error. The proposed project seeks to examine the feasibility of a bold and ambitious undertaking which, if successful, would solve this and many kindred problems and open up an area of considerable research and development potential. Phase I research will focus on proving the feasibility of the idea of obtaining a set of measurement data from an animal (specifically, the horse) at a normal walking pace and adequate navigational tools to allow the user to sift through the 3D video streams to make the adjustments needed to obtain a fully tracked model.The current methods for the static study and measure of humans and animals is either the low-tech tape or the full-body laser or active-light scanner that requires a large booth. Even after obtaining the whole-body scan, as with the clothing industry, obtaining the desired measurements to use in applications is difficult because of a lack of convenient landmarks. Landmarks are made visible through motion. The goal is to make use of motion and its cyclical aspects to identify the landmarks and to recover, not just a single-position 'statue' of the subject, but its entire motion. In the U.S., there are around 10 million dairy cattle, of which about 34%, or 3.4, million are measured every year by tape or by little more than guessing. A capability to obtain measurements fast and accurately would have ready use in the dairy industry as well as in the thoroughbred industry.