Introduction

After initial processing by a Digital Signal Processor (DSP), the reduced data rate will fit through an asynchronous serial port at $115\,kb/s$ for full analysis and tracking in the host computer. The host must ...

• reliably retrieve the streaming data from the serial port
• filter and scale the data according to factory calibration data
• analyze the data for object parametrics, position and orientation
• translate the position into user-calibrated video alignment
• mark the threatening weapon's position on the camera image

Some frames, grabbed from the video stream during a demonstration, are shown in figure 1. The demonstrator was holding the gun in his right hand. The blob obscures the actual position of the gun and the red color indicates that the magnetic signature matches a dangerous object category.

For details on the underlying technologies, readers are referred to the literature on magnetic tensor gradiometry[1], magnetic position extraction[2], threat identification[3] and other applications[4]. This paper only summarizes the algorithms to show the computational impact of their numerical implementations.

The software development for this project was initially conducted using Microsoft tools to take advantage of existing source code. The prior project recorded raw data without any calculation and subsequently used modelling software for analysis. The central part of this paper reviews the new Linux-based implementation for this project, including descriptions of the problems that were encountered and shows how each problem was resolved as the code base for the project was migrated to Debian GNU/Linux.