omxipcs_ycbcr420tobgr888ls_mcu_s16_u8_p3c3.c

The OpenMAX DL (Development Layer) APIs contain a comprehensive set of audio, video, signal processi

/**
 *
 * 
 * File Name:  omxIPCS_YCbCr420ToBGR888LS_MCU_S16_U8_P3C3.c
 * OpenMAX DL: v1.0.2
 * Revision:   10586
 * Date:       Wednesday, March 5, 2008
 * 
 * (c) Copyright 2007-2008 ARM Limited. All Rights Reserved.
 * 
 * 
 *
 * Description  : Colour Conversion Routine - Converts YUV420 to BGR888 in floating point.
 *                Source data is stored as a JPEG MCU in three channel, planar domain and 
 *                the destination would be in three-channel pixel format.
 *
 */

#include "omxtypes.h"
#include "armCOMM.h"
#include "armOMX.h"
#include "omxIP.h"
#include "armIP.h"

/**
 * Function:  omxIPCS_YCbCr420ToBGR888LS_MCU_S16_U8_P3C3   (4.4.3.8.3)
 *
 * Description:
 * These functions convert sub-sampled YCbCr data to BGR888 data with 
 * level-shift.  The parameter pSrcMCU[0] must point to the start of a 
 * sufficient number of contiguously stored 8x8 Y blocks to constitute a 
 * complete MCU given the source chroma sub-sampling scheme, i.e., 1, 2, or 4 
 * blocks for YCbCr 4:4:4, 4:2:2, or YCbCr 4:2:0, respectively. 
 *
 * Input Arguments:
 *   
 *   pSrcMCU - buffer containing input MCU pointers:  The parameter 
 *            pSrcMCU[0] points to the first Y block, pSrcMCU[1] points to Cb 
 *            the block, and pSrcMCU[2] points to the Cr block; all three 
 *            pointers must be aligned on 8-byte boundaries.  The parameter 
 *            pSrcMCU[0] must point to the start of a sufficient number of 
 *            contiguously stored 8x8 Y blocks to constitute a complete MCU 
 *            given the source chroma sub-sampling scheme, i.e., 1, 2, or 4 
 *            blocks for YCbCr 4:4:4, 4:2:2, or YCbCr 4:2:0, respectively.  
 *            Only top-down storage format is supported.  Input components are 
 *            expressed using a Q8 representation and are bounded on the 
 *            interval [-128, 127]. 
 *   dstStep - distance in bytes between the starts of consecutive lines in 
 *            the destination image; values less than 0 are allowed to support 
 *            bottom-up storage format. 
 *
 * Output Arguments:
 *   
 *   pDst - points to the output image buffer in which the output data (C3 
 *            representation) are interleaved as follows:  BGRBGRBGR...  
 *            Bottom-up storage format is supported. Outputs are saturated to 
 *            the OMX_U8 range [0, 255].  The parameter dstStep can take 
 *            negative values to support bottom-up storage format. 
 *
 * Return Value:
 *    
 *    OMX_Sts_NoErr - no error 
 *    OMX_Sts_BadArgErr - bad arguments.  Returned if one or more of the 
 *              following was true: 
 *    -   a pointer was NULL 
 *    -   the absolute value of dstStep was smaller than 24 (for YCbCr444) 
 *        or 48 (for YCbCr422/YCbCr420) 
 *    -   a pointer in pSrcMCU[] was not 8-byte aligned 
 *
 */

OMXResult omxIPCS_YCbCr420ToBGR888LS_MCU_S16_U8_P3C3(
    const OMX_S16 *pSrcMCU[3],
    OMX_U8 *pDst,
    OMX_INT dstStep
)
{   
    const OMX_S16 *pSrcY, *pSrcY2, *pSrcCb, *pSrcCr;
    OMX_S16 Ydata, Ydata2, Ydata3, Ydata4, Udata, Vdata;
    OMX_INT i, j, blk_x, blk_y;
    ARM_BLOCK8x8 *y8x8;
    OMX_INT blkOffset_Dst, blkOffset_UV;
    OMX_U8  *pRGB, *pRGB2;
    ARM_JpegPixelYUV2RGBConversionFunctionType pixCCfn = armIPCS_ComputeRGBFromYUV_JPEG_F32_U8;
    
    armRetArgErrIf(!pSrcMCU, OMX_Sts_BadArgErr);
    armRetArgErrIf(!pDst, OMX_Sts_BadArgErr);
    armRetArgErrIf(!pSrcMCU[0], OMX_Sts_BadArgErr);
    armRetArgErrIf(!pSrcMCU[1], OMX_Sts_BadArgErr);
    armRetArgErrIf(!pSrcMCU[2], OMX_Sts_BadArgErr);
    armRetArgErrIf(!armIs8ByteAligned(pSrcMCU[0]), OMX_Sts_BadArgErr);
    armRetArgErrIf(!armIs8ByteAligned(pSrcMCU[1]), OMX_Sts_BadArgErr);
    armRetArgErrIf(!armIs8ByteAligned(pSrcMCU[2]), OMX_Sts_BadArgErr);
    armRetArgErrIf(((dstStep > (-3 * JPEG_MCU_PIX_WIDTH_420)) && (dstStep < (3 * JPEG_MCU_PIX_WIDTH_420))), OMX_Sts_BadArgErr);

    y8x8 = (ARM_BLOCK8x8 *)pSrcMCU[0];
    
    for(blk_y = 0; blk_y < 2; blk_y++)
    {
        for(blk_x = 0; blk_x < 2; blk_x++)
        {
            blkOffset_Dst = blk_y * 8 * dstStep + blk_x * 24;
            blkOffset_UV  = blk_y * 8 * 4 + blk_x * 4;
            
            pSrcY   = (OMX_S16 *)y8x8++;
            pSrcY2  = pSrcY + 8;
            pSrcCb  = pSrcMCU[1] + blkOffset_UV;
            pSrcCr  = pSrcMCU[2] + blkOffset_UV;
            pRGB    = pDst + blkOffset_Dst;
            pRGB2   = pRGB + dstStep;
            for(i = 0; i < 4; i++)
            {
                for(j = 0; j < 4*6; j+=6)
                {
                    Ydata       = *pSrcY++;
                    Ydata2      = *pSrcY++;
                    Ydata3      = *pSrcY2++;
                    Ydata4      = *pSrcY2++;
                    Udata       = *pSrcCb++;
                    Vdata       = *pSrcCr++;
                    
                    pixCCfn(Ydata, Udata, Vdata, &pRGB[j+2], &pRGB[j+1], &pRGB[j]);
                    pixCCfn(Ydata2, Udata, Vdata, &pRGB[j+5], &pRGB[j+4], &pRGB[j+3]);
                    pixCCfn(Ydata3, Udata, Vdata, &pRGB2[j+2], &pRGB2[j+1], &pRGB2[j]);
                    pixCCfn(Ydata4, Udata, Vdata, &pRGB2[j+5], &pRGB2[j+4], &pRGB2[j+3]);
                }
                pSrcY   += 8;
                pSrcY2  += 8;
                pSrcCb  += 4;
                pSrcCr  += 4;
                pRGB    += 2 * dstStep;
                pRGB2   += 2 * dstStep;
            }
        }
    }
    return OMX_Sts_NoErr;
}

/* End of file */