About 10-bit pixel datum from CMOS image sensor

Hi all,

I am sorry if this problem has been answered but I cannot find much help from the archive.

I am using a Spartan3 starter board to interface with a Micron CMOS Image sensor (MT9T001), and is confused with the output data from the sensor.

It outputs 10-bit pixel datum in each pixel clock, but I am wondering what is the use of the last 2 bits?

From my understanding, each pixel datum is of 0~255 value (2-to-power-8).

Thanks!

Nick

Reply to
Nick
Loading thread data ...

Nick, Try to search for Camera Link Protocol on the web and you will find answers to your question. Such topics have already been discussed on this forum as well.

Hope this helps

/MH

Reply to
mh

MH,

Thanks! I search that protocol as you suggested but Micron's CMOS image sensor seems not to be registered, and it is not following that protocol.

Nick

mh wrote:

Reply to
Nick

Hi, with 10 bits pixel you have more accuracy for each pixel. If you need only 8 bit for each pixel, just ignore the 2 LSBs.

Matteo

Reply to
Teo

Nope - each pixel is of value 0-1023. It makes it difficult to display what the FPGA can process - there are details that it will pick up that you can't see on the screen (I'd be surprised if your monitor truely has 256 levels of gray you can distinguish anyway :-)

It makes data storage a bit more hassle as well, not many formats support 10 bit data and even fewer packages import it. imageMagick is one that can AFAIK, and PGM files can hold 10 bit data.

We've seen the benefits of using the bits that are there though!

Cheers, Martin

--
martin.j.thompson@trw.com 
TRW Conekt - Consultancy in Engineering, Knowledge and Technology
http://www.conekt.net/electronics.html
Reply to
Martin Thompson

Technology

formatting link

You can also use the 10 bits to get a little more dynamic range from the camera even if you only display with 8-bit resolution. I have used two approaches to this. One is to set the "digital gain" parameter to 1, 2, or 4. This "gain" selects whether you see the upper 8 bits, the middle 8 bits, or the lower 8 bits respectively. Usually when selecting gain 2 or 4, the FPGA would saturate the pixels where the unused upper bits are not zero.

Another approach is a quasi-logarithmic function that translates

10 bits to 8 bits. This is useful for visual images that don't require linearity. It allows you to see some contrast in otherwise dark regions of an image.

Regards, Gabor

Reply to
Gabor

Also worth noting that the SMPTE-292 spec is for 10 bit video, so while the SDI standards are generally 8 bit, some of the HD standards have 10 bit.

Reply to
Ray Andraka

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.