Video/Image Processing on FPGA OR Micro-controller?

Hello Friends,

Seems like a cool forum... I think this is the best place to ask this questions.

We design Thermal Imaging cameras which need quit a allot of custom image processing and custom sensor controls. What should we use FPGA or Micro controller?

Camera and video/image processing Brief

We have a 1024 by 768 12 bit per pixel(BW) image 30fp , coming in through a serial link.

The image then needs allot of image processing such as :

3 point Calibration Custom Contrast enhance and detail enhancement Scaling Colourization AGC Blending Object detection

FPGA Approach: Completely possible no problem with custom device interfaces and horsepower.... However its an expensive approach. Has longer development time and seems to more difficult.

I am very interested in be able to use on of the TI MC such as Omap4.

This seems very appealing because of its onchip interfaces:

Memory Controllers Serial /Usb controllers Wifi/3g capabilities The device is composed of the following subsystems: ? Cortex?-A9 microprocessor unit (MPU) subsystem, including two ARM® Cortex-A9 cores

? Digital signal processor (DSP) subsystem ? Image and video accelerator high-definition (IVA-HD) subsystem ? Cortex?-M3 MPU subsystem, including two ARM Cortex-M3 microprocessors

? Display subsystem ? Audio back-end (ABE) subsystem ? Imaging subsystem (ISS), consisting of image signal processor (ISP) and still image coprocessor

(SIMCOP) block ? 2D/3D graphic accelerator (SGX) subsystem ? Emulation (EMU) subsystem

However as I have not worked much with these kind of multimedia application processor. I have not understood whether this kind of processor can be used for my application.

Can I implement my custom image processing blocks using the on chip DSP Cortex?-A9 microprocessor unit (MPU) subsystem ,Image and Video Acceleration and ISS blocks.

Is it possible to implement such processing algorithms:

3 point Calibration Custom Contrast enhance and detail enhancement Scaling Colourization AGC Blending Object detection

and then Output it on an LCD.

Look forward to hear from you all soon.



--------------------------------------- Posted through

formatting link

Reply to
Loading thread data ...


You have a 284Mbps serial link? What kind?

Reply to

This is a newsgroup and your use of the word 'cool' immediately makes me think a university project.

You happened to see a copy of the newsgroup on a website.


Sounds pretty typical requirements. Mixture of point and area processing.


How do you KNOW this?

If from past experience you should have parts of previous FPGA designs that can be reused or re-purposed for such things, almost like resusing parts of previous circuits.

You mean you want to play with software and a device that is new to you.

There is the THIRD possibility FPGA and processor, where you do as much time critical path in the FPGA and less time critical in processor with wome FPGA paths controlled by processor (changing calibration points, AGC range, colorisation factors etc..). FPGA and LUTs inside them are very good for offloading point processing to minimal delay. Even convolution processing becomes relatively easy with a few line delays in FPGA

That way you get less missed frames due to other issues of Wifi protocol taking a lot of CPU time and dropping frames.

Yes a datasheet feature list, if you have not used them before you could well be taking a longer development time getting each stage working.

The first question you have to ask is -

"What is the maximum end to end frame delay the system can have under ANY circumstances and external influences".

When you know that then you can work out what happens in the SYSTEM.

Then decide which parts to use or combinations.


A lot of multimedia applications do not care about occassional drop frmaes or total end to end delay. So if playing from a DVD and it takes

10 or 20 frame delays from reading from DVD to display it does not matter as theer is a consistent delay.

If this camera is used by an operator and has a noticeable LAG when moving the camera around it will become difficult to use, if not cursed.

Insufficient information on the SYSTEM requirements to detrmine this.

Insufficient information on product lifetime, cost, or physical size to determine which solution is best.

Insufficient information on your capabilities to determine timescales or how easily you will be able to do this.

Paul Carpenter          |
    PC Services
 Click to see the full signature
Reply to

If you have never done FPGA-based design, this is too ambitious a project to start with.

--------------------------------------- Posted through

formatting link

Reply to

Stupident go ask your prof.

njamal wrote:


gh a


) subsystem, including two ARM=E2=94=AC=D0=BE


two ARM Cortex-M3

processor (ISP) and



and Video

Reply to
Vladimir Vassilevsky

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.