7 years, 5 months ago.

JCU module errors

Is it possible to make JCU module tolerate reset marker errors as (libjpeg) does. For example libjpeg says "Corrupt JPEG data: 1 extraneous bytes before marker 0xd3" and even though the JPEG is successfully decoded and viewable. However if you try to decode the same image with the JCU it just exits with error "JPEG_CONV_JCU_ERR". Аs far as I read the hardware manual of the RZ/A1H there isn't such possibility and in this connection ... Does anyone knows a decent way for transcoding from MJPEG to JPEG so the JCU module do not complain. (The source MJPEG stream is from a Logitech C270 web cam. Should I try another brand ?)

Question relating to:

GR-PEACH is an Mbed enabled platform which combines the advantages of the Mbed ecosystem and Arduino UNO form factor.

2 Answers

7 years, 5 months ago.

I think that it can be realized by changing ErrorFilter with "R_JCU_SetErrorFilter()".

#include "JPEG_Converter.h"
#include "r_jcu_api.h"

{
    JPEG_Converter  decoder;
    JPEG_Converter::bitmap_buff_info_t  aBitmapData;

    aBitmapData.width           = SAMPLE_WIDTH;
    aBitmapData.height          = SAMPLE_HEIGHT;
    aBitmapData.format          = JPEG_Converter::WR_RD_YCbCr422;
    aBitmapData.buffer_address  = (void *)JCUBuffer_OUTPUT;

    // Set Error Filter
    R_JCU_SetErrorFilter(JCU_INT_ERROR_SEGMENT_TOTAL_DATA | JCU_INT_ERROR_MCU_BLOCK_DATA);

    // JPEG_Converter
    decoder.decode((void *)JCUBuffer_INPUT, &aBitmapData);
}

Accepted Answer

Yes this actually do the trick, the module does not complain anymore, but the result image is still corrupted (there are lots of artefacts). Anyway, may be I just should put up with that and use the software libjpeg. Thank you !

posted by Vekatech Ltd. 18 Nov 2016
7 years, 5 months ago.

The recommended CMOS Micron economical camera is MT9V111 which is designed to work with GR-PEACH. And is available from digi-key with audio video shield. The GR-PEACH WebCamera application and QR barcode ZX application also uses this camera to read and convert 2d QR codes into text ASCII. Other video in is via Audio video NTSC shield board. Manual focus is available on MT9V111 camera module.