Implementation of Correlation Filter on DSP Processor for Image Based Applications

In this century, computer vision has a wide range of applications in various fields of life. The mostly used application is the object detection. Automatic target detection can be done using different types of filtering techniques. This paper describes the filtering algorithm for detecting the targe

Project Title

Implementation of Correlation Filter on DSP Processor for Image Based Applications

Project Area of Specialization

Computer Science

Project Summary

In this century, computer vision has a wide range of applications in various fields of life. The mostly used application is the object detection. Automatic target detection can be done using different types of filtering techniques. This paper describes the filtering algorithm for detecting the targets.  This technique contains the MACH filter. We are looking for the working and implementation of MACH filter on Digital Signal Processor. Simulation and implementation on DSP provides the comparative result. This implementation is done on the Digital Signal Processor (DSP) kit TMS320C6713 for Target detection using correlation filters on Embedded platform for image processing applications leading to future enhancement for real-time applications. The developed system will be able to not only assist in security of civilians but also will aid the armed forces in foe detection. 

Project Objectives

The objectives of this project are:

• Enhanced target detection using correlation filters on embedded platform such as DSP for image processing applications leading to future enhancement for real-time applications
• Enhanced security of civilians in crowded places
• Enhanced security of armed forces personel

Project Implementation Method

The MACH filter:

The MACH filter maximises the correlation peak when the target is detected. It can also be called Optimal Tradeoff MACH (OT-MACH) filter because its characteristics vary according to the requirements.

The general properties of OT-MACH filter are that it:

• Maximizes the Average Correlation Height (ACH)
• Reduces the Output Noise Variance (ONV)
• Balances the Average Similarity Measures (ASM)
• Also maintains Average Correlation Energy (ACE)

The basic energy equation of MACH filter is given by:

Where, Where, ?, ?, and ? are the OT Parameters which range from 0 to 1.

On reconstructing the equation, we have our desired MACH filter:

Where, m is the average of training images in frequency domain and * denotes the conjugate of the image. C is the power spectral density matrix representing additive input noise as:

Where,  is the noise variance and I is the identity matrix. Dx is the diagonal average power spectral density of the training images i.e.

Where, Xi is a matrix of training images. Sx denotes the similarity matrix of the training images and M is the average of all the Xi:

TMS320C6713:

The hardware used for the implementation of MACH filter is the DSP Starter Kit (DSK) TMS320C6713 designed by Texas Instruments. Code Composer Studio incorporates a C compiler, an assembler, and a linker to generate C6x executable files. The generic block diagram of DSK is given below:

The DSK features the TMS320C6713 DSP, a 225 MHz device delivering up to 1800 million instructions per second (MIPs) and 1350 MFLOPS.  This DSP generation is designed for applications that require high precision accuracy. The C6713 is based on the TMS320C6000 DSP platform designed to needs of high-performing high-precision applications such as pro-audio, medical and diagnostic. Other hardware features of the TMS320C6713 DSK board include:

• Embedded JTAG support via USB
• High-quality 24-bit stereo codec
• Four 3.5mm audio jacks for microphone, line in, speaker and line out
• 512K words of Flash and 16 MB SDRAM
• Expansion port connector for plug-in modules
• On-board standard IEEE JTAG interface
• +5V universal power supply

Work Flow:

The basic block diagram of work is given below:

The work flow goes through the following steps:

1. Training data images are read in MATLAB.
2. MACH filter design is done in DSK C6713.
3. Testing image is read in MATLAB and sent to DSK.
4. MACH filter is applied on the image.
5. Result is received back in MATLAB and output is shown.

Benefits of the Project

The benefits of this project are:

• Target detection can be efficiently done in defence and security applications
• Computational time can be minimised using DSP kit in compared with computer system
• A portable embedded system can be provided

Technical Details of Final Deliverable

The final deliverable will provide the following services:

• Picture of the target will be captured in realtime
• The picture will be matched with the pre-trained MACH filter
• The result will tell us that either the target has been detected or not.

Final Deliverable of the Project

HW/SW integrated system

Core Industry

Security

Other Industries

IT , Security , Telecommunication

Core Technology

Big Data

Other Technologies

Others

Sustainable Development Goals

Decent Work and Economic Growth, Industry, Innovation and Infrastructure, Sustainable Cities and Communities

Required Resources

If you need this project, please contact me on contact@adikhanofficial.com