Introduction

Enhanced the Parcel Scanning research result for Melbourne University and Collaboration with Melbourne University Consortium, Extel Technologies and LSI. The Parcel Scanning research hypothesize into the Artificial Intelligence by training the digital scan to detect organic material to be trained to detect drugs and metal material to be trained to detect gun and explosion. The enhancement comprised of turning the research result into the commercial product including proving the result was correct with better technique

Project Team

  • Duration ~6 months
  • 1 x System Architect ~ Extel Technologies CTO was responsible
  • 1 x Project Manager/Tester ~ LSI Subject Matter Expert and Melbourne University research for the scanning and digital signal processing
    • Document the research result
    • CTO and SME provide guidance to the Senior Software Engineer and Developer
  • 1 x Senior Software Engineer and Developer
    • Work under the guidance of the Extel System Architect and LSI SME
    • Reverse engineering and undo all necessary components of the current solution and enhance the research result each day and each week under the guidance of the System Architect and SME whenever possible
    • Turn the research to the product by systematically separate the Image Processing Libraries and the LSI Scanning Application from the Linux Operating System
    • The LSI Application and the Libraries are running as part of the Linux Account
    • Provide all graph and plot for the organic and metal against all the reference/training material. As the result of the training

Technologies

Reverse Engineering

  • As there was no documentation, we created a macro to print. We ran the application to study the concurrency nature of the digital signalling processing and isolating all the experiment implementation
  • We also attempted to use the Rational Rose to document class diagrams and sequence diagram to understand the software architecture of the current research application
  • The server was a powerful server with 4CPU/8Core per CPU, 128GB of Memory and Linux Centos 6 (closet to the commercial Red Hat)
  • We removed all root permission to the application by
    • We setup the DevEnv using Jetbrains CLion IDE running on Linux
    • We updated the cmake to latest and ensure the all the libraries are compile and install to the lsi include files and libraries
    • Create lsi account and installing and configuring all the libraries to the lis account
    • We ran the application with all the necessary hash define (#define/#ifdef) to print and turn-on/off the printing
  • Once we had the Dev-Env established:
    • We replayed all the data captured to understand the code coverage and concurrency with threading
    • We ensured that all memory and threads are fully utilised and there is no memory leakage on every scan.
    • We applied various Fast Fourier Transform techniques to prove the current application
    • We applied graph and plotting using OpenCV and VTK libraries
    • We trained the LSI scanner to understand the reference organic and metal and other interested materials.
    • We acquired and compare the result by graphing and plotting the real time scanning result against the reference result.

Looking back: We wonder about the achievement: “Given that I did not touch the development for more than 10 years. I was put into a very deep and dark end of the knowledge universe where the application contained no documentation, the research was a mixture solutions and techniques developed by various researchers and software developers and I was still being able to develop and deploy the solution under the guidance of the Extel CTO and LSI SME – Special thank to the Extel CTO and LSI SME to provide all the guidance on the Digital Signal Processing techniques and guidance each day. By the way we shorted the development planned for 6 months down to 4 months.”