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.

TDA2PXEVM: Alg_RgbIRProcess Documentation

Part Number: TDA2PXEVM

We have just found there is an usecase for TDA2px with RGB-IR input implemented (and compiled for DSP C66x), Do you know where can we find more information about the plugin configuration (such as, input Bayer pattern, Image Size, and output format, and the parameters meaning):

  

    /* Set Parameters for new buffers */

    pUcObj->Alg_RgbIRProcessPrm.maxOutputHeight  = 2048;

    pUcObj->Alg_RgbIRProcessPrm.maxOutputWidth   = 4096;

    pUcObj->Alg_RgbIRProcessPrm.maxInputHeight   = 2048;

    pUcObj->Alg_RgbIRProcessPrm.maxInputWidth    = 4096;

    pUcObj->Alg_RgbIRProcessPrm.numOutBuffers    = 4;

 

    pUcObj->Alg_RgbIRMuxPrm.maxOutputHeight      = 2048;

    pUcObj->Alg_RgbIRMuxPrm.maxOutputWidth       = 4096;

    pUcObj->Alg_RgbIRMuxPrm.maxInputHeight       = 2048;

    pUcObj->Alg_RgbIRMuxPrm.maxInputWidth        = 4096;

    pUcObj->Alg_RgbIRMuxPrm.numOutBuffers        = 4;

 

   pUcObj->Alg_RgbIRProcessPrm.gainAdj = 16;

   pUcObj->Alg_RgbIRProcessPrm.dc      = 168;

   pUcObj->Alg_RgbIRProcessPrm.alpha   = 128; //Q7

   pUcObj->Alg_RgbIRProcessPrm.rgbThr  = 4000;

   pUcObj->Alg_RgbIRProcessPrm.irThr   = 2000;

   pUcObj->Alg_RgbIRProcessPrm.lowThr  = 512 ;

   pUcObj->Alg_RgbIRProcessPrm.Mode    = 1;  /* 0: Day, 1: Night */

   pUcObj->Alg_RgbIRProcessPrm.IRModeGain[0]    = 128;  /* R gain Q7 */

   pUcObj->Alg_RgbIRProcessPrm.IRModeGain[1]    = 128;  /* G gain Q7 */

   pUcObj->Alg_RgbIRProcessPrm.IRModeGain[2]    = 128;  /* B gain Q7 */

   pUcObj->Alg_RgbIRProcessPrm.IRModeGain[3]    = 128;  /* IR gain Q7 */

 

 

Can you confirm the RGB-IR plugin is compatible with this 4x4 RGB-IR pattern:

 

Can you provide us a high level point of view what is the current scope for this RGB-IR implementation? 

Thanks in advance, 

Best regards,

Andrés

  • Hi,

    >>> We have just found there is an usecase for TDA2px with RGB-IR input implemented (and compiled for DSP C66x)

    Can you point to the source code location? 

    Thanks

    Gaviraju

  • The source code is here:

    C:\PROCESSOR_SDK_VISION_03_08_00_00\vision_sdk\apps\src\rtos\alg_plugins\rgbirprocess\

  • The source code for RGB-IR settings is here:

    C:\PROCESSOR_SDK_VISION_03_08_00_00\vision_sdk\apps\src\rtos\usecases\iss_capture_rgbir_isp_display\chains_issRgbIrIsp_Display.c

  • Hi,

    We don't have a document specific to the RGBIR plugin implementation & in every plugin we can see the create function, where we are configuring the I/P & O/P channel settings based on the previous channel info.

    Please refer to the create function of the RGB-IR algorithm plugin to know more information about I/P & O/P channel settings.

    I'm not sure this algorithm plugin support 2x2 or 4x4 RGB-IR pattern, I will check with an expert & let you know.

    Thanks

    Gaviraju

  • Hello Gaviraju,

    Thanks for your reply. I am working with the OV2312 module camera from Leopard Imaging (over FPDLINK III), and I dont know if this plugins would be fine for my purpose, or maybe I should do a own implementation.

    Can you provide more information about this plugins capabilities? would it be viable adapt this plugin for OV2312?

    BR,

  • Hello Andres,

    The usecase was tested with AR0239 which is 2MP (1920x1080) sensor. RGBIR CFA is 4x4.

    Parameters like maxOutputHeight  , maxOutputWidth are used for memory allocation for worst case. Typical output resolution is same as sensor resolution if there is no scaling.

    Other parameters are used to preprocess RGB-IR and separate them into RGB and IR buffers. For e.g.

       pUcObj->Alg_RgbIRProcessPrm.dc      = 168; This is the DC pedestal of AR0239.

       pUcObj->Alg_RgbIRProcessPrm.alpha   = 128; //Q7 This is the coefficient used for IR cancellation i.e. subtracting IR from RGB to get true black.

     

    To understand all the parameters you will need to go through algorithm details.