This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

TDA3XEVM: Can Image Signal Processor(ISP) Process 14-bit monochrome video?

Part Number: TDA3XEVM
Other Parts Discussed in Thread: DM505

Our image sensor is Vanadium Oxide(VOx) Resistive Microbolometers Infrared Thermal Detectors Focal Plane Array(FPA).

Our application is night vision.  It's a hunting application.  Our product is RifleScope.

FPA output is 14-bit, 640 x 480, monochrome, High Dynamic Range, it's not companded.  We can select 12-bit monochrome mode in TI ISP for processing 14-bit image sensor output.  Do we have any options with TI chip to process all 14 bits?

Thanks in advance.

Regards,

Amer

  • Hi Amer,

    can you check if the suggestions in this thread can help you:
    e2e.ti.com/.../2769641

    Regards,
    Yordan
  • The thread is very useful but to utilize it I need to understand it.

    TI says, Option 1 - ISSM2MISP_LINK_OPMODE_2PASS_WDR - You will need to specify decompanding parameters for 14->20 bit. ISP will split it into two frames of 12-bits each, process them separately and merge at the end.  Does this option apply on when sensor output is companded?

    TI says Option 2 - ISSM2MISP_LINK_OPMODE_1PASS_WDR - You will need to specify a Global Tone Map table to convert 14-bit data into 12-bit data.
    Does this option apply to my sensor?  If yes, specifying global tone map means compressing to 12-bit?  Then, will ISP will uncompress it back to 14-bit HDR?  What algorithm do I use to convert?  Do I divide by 4
    and quantize the 16,384 levels to 4,096 levels?  How will ISP know the data was converted from 14-bit ?

    Regards,

    Amer

  • hi Amer,

    Please find answers below,

    1, yes, it can be applied for the companded sensor output.

    2, global tone mapping can be done using two method, PWL and LUT (curve defined by lut) based mapping.  14 bit data cannot be processed in ISP, so you will need to tone map 14 bit data to 12bit.

    Rgds,

    Brijesh

  • Option 1 - Our image sensor output will be raw, monochrome, 14-bit, WDR, 640 x 480. Image sensor will be connected to FPGA. After processing option 1 or option 2 in FPGA, data will be presented to TI DM505 ISP.
    TI suggests "ISSM2MISP_LINK_OPMODE_2PASS_WDR - You will need to specify decompanding parameters for 14->20 bit. ISP will split it into two frames of 12-bits each, process them separately and merge at the end."
    "specify decompanding parameters for 14->20 bit" Please explain this phrase so person with little background in Image Signal Processing can understand. By decompanding, do you mean decompress? Do I compress in FPGA and specify decompanding parameters so ISP can determine what the pixel values of the raw 14-bit video are?
    "14->20 bit" what does this exactly mean?
    Is this option equivalent to processing all 14 bits of raw video?
    Please provide page numbers to read in TRM and other resources like books, etc. so we can implement this option with confidence.

    Option 2 - TI suggests "global tone mapping can be done using two method, PWL and LUT (curve defined by lut) based mapping."
    What is PWL stand for? We'll do this in FPGA and output result to ISP. Please provide resources that explain these two methods, TRM pages, books etc.
    Is this option equivalent to processing all 14 bits of raw video?

    Thank you and Regards,
  • 14->20 decompanding is needed only if the sensor provides 20bit WDR images companded to 14 bits. In your case, the sensor is probably generating 14-bit linear images. You need to just apply a LUT tone map at IPIPEIF stage to convert it to 12-bit.

    This step is very similar to gamma compression.