The Design And Implementation Of Visual Multi-Target Positioning System

Screen positioning is widely used on whiteboards, tablet PCs and other devices, users cantouch and write with a finger, which makes the operation more convenient and intuitive thanbefore. However, most of existing positioning technology can only achieve single-touch,and lack intelligence and naturalness. Multi-point positioning breaks this limitation, itmakes multi-touch and multi-user concurrent operation possible. As an efficient interaction,multi-touch can provide more innovative experience, and is now attracting more and moreattention, it will certainly have great potential and broad prospects in the future.This paper researchs the multi-target positioning technology and implement a system.This system captures and processes infrared images, then detectes targets, finally uses amodel of multi-target positioning to calculate the coordinates.The main work is as follows:(1) Design the overall project of the system and build the platform of hardware. Weconstruct the infrared environment and configure the sensor, design the transmission modulebased on CY7C68013A and develop the firmware, configure the boot file of USB driver andfinally drive hardware platform to work;(2) Process the captured images, including background subtraction, mean filtering, andthreshold segmentation. Design a method of multi-target detection to detect multi-targetrapidly;(3) Research the multi-target positioning and tracking algorithm. We divide cameras intogroups and employ a minimum distance method to track multiple targets. The look-up tablemethod is employed to reduce nonlinear distortion of the lens;(4) Develop the program of up-lever computer.The program achieves device recognition,image acquisition and transmission, image processing, target detection, coordinatescalculation, targets tracking and other functions.We also develop a drawing program to testand analyze the performance of the system.The main innovations are as follows:(1) Propose a multi-target detection method. This method uses one-dimensional arraysinstead of processing a large number of data to detect multi-target. By choosing appropriatethresholds, it can achieve the rapid detection of multiple targets, with a small computationalload, short response time, and low hardware requirement, it is suitable for system demandshigh real-time;(2) Propose a multi-target positioning method. The method divides multiple cameras into groups, they shoot from multiple angles, targets are determined by data from intersectionmeasurement. This method can effectively cope with occlusion and eliminate false targets.The results of test shows that the system can position multiple targets, with a simplestructure, low cost, high accuracy, without size limitations.It provides a good solution fortouch screen, and has a certain application and reference value.