A Study On Depth Imaging Technology For Natural Human-Computer Interaction

With the rapid development and popularization of computer technology, computer morphological types are continuously enriched. People put forward more and more demands on the friendly and useful human-computer interaction. As an important input technology of natural human-computer interaction, depth-imaging technology becomes more and more concerned. In order to adapt to the mobile terminal and wearable devices, how to make the cost of depth-imaging computation smaller and the equipment of it easier, has become the focus of our research.In this paper, a monocular depth imaging system based on structured light and infrared intensity is established for the near range (20-60cm) natural human-computer interaction. The system is mainly composed of three parts:the timely switched on structure light part, the infrared part with adjustable illumination, and the single CMOS image acquisition part. In order to reduce the complexity both in time and memory, we employ a two-stepping method to match the near infrared intensity image to absolute depth. A simple pattern structured light is used to classify the infrared image into coarsely quantized depth ranges (5cm) at the first step. Then, the second step, we estimate the absolute depth value by intensity differences from near-infrared imaging. Experiments show that at the depth of 46cm, our depth imaging system can distinguish the depth difference of 5mm, and basically reach the accuracy requirement of 1%.Firstly, the paper introduces the development of human-computer interaction and the application of natural human-computer interaction in the world, and then summarizes the principle and researches of the main depth-imaging technique. After analyzed the characteristics of near range natural human-computer interaction scenario and its requirements for depth-imaging system, we described our two-stepping depth-imaging method. Then from the image acquisition module, structured light module, and infrared illumination module, a concrete analysis of the system accuracy realization lead to the specific device selection and opto-mechanical system design. At last, we established the training and the accuracy testing experimental system, and tested our depth-imaging method by experiments.