Detection and targeting of moving objects using digital optical stereo imaging technique

The main concern in the present work is to determine the position of far moving object/objects. From the coordinates of this position the depth, or the distance, of an object to observers in any known position in the real-world space could be determined. The method we have utilized depends mainly on how to use the stereo technique to achieve the goal. The procedure we have developed to detect such objects consists of two main software systems. The first part of procedure constitutes the software required, by each terminal in the system, for detecting the location of the moving objects in a frame (image at time 1). Furthermore, it concerns with finding some information about the objects' shapes by looking at them by each terminal from
different viewing angles. Also, this part of system converts the image to binary, removes noise, filling the object's holes, detects edges of the objects in that image, and finally starts a tracking procedure that extracts some information concerning the geometrical features of the detected object, as soon as an edge is detected. A new edge detector Thinning Edge Detector (TED) is developed in the present work. We have demonstrated how TED is capable of producing connected one-pixel wide edges. The second part of the developed system is a software stereo system that deals with the information supplied to the main station by the terminal stations or channels. From the supplied information, object's location in the real-world coordinates in each pair of frames are computed. The chosen rate of imaging, in this study, is five frames per second. Accordingly, we have calculated the displacement, velocity, and acceleration of the moving object. To test the efficiency of the developed system, real and simulated experiments '"- were performed. The results of the real experiments performed were limited due to the lack of real stereo imaging system. The simulation experiments (using 3D-MAX '.package) have been performed so many times, for many different stereo cases, in the \same environment. The immunity of the system towards three types of additive noise has been tested. The added noise were Gaussian, uniform, and impulsive. The system proved it is very immune to the first two types of noise even if they present with high probability, while its immunity to the third is high only with moderate probabilities. All the programs, in the framework of the present study, were written in language and executed by using a 486 IBM, and all the simulated images that Produced by the 3D-MAX package, have been achieved with the use of a Compaq pentium-II) computer.