Converting yuv 4:2:0 file to Rgb not getting expected output - c++

Question

The logic I am using is that I am reading he yuv file in a buffer and using 3 pointers to point to the Y,U,V components. The image dimension is 1920*1080.
I am
taking 4 y pixels for corresponding 1U,1V pixels.
Extracting the pixel value in integers.
Converting the Y,U,V to R,G,B and storing the components in RGB buffer.
But the video output is incorrect. The output is somewhat like black and white
#include "stdafx.h"
#include <string>
#include <stdio.h>
unsigned char* g_pcRGBbuffer;
unsigned char* g_pcYUVBuffer;
int _tmain(int argc, _TCHAR* argv[])
{
int l_nSize = 1920 * 1080 * 1.5;
g_pcYUVBuffer = new unsigned char[l_nSize];
g_pcRGBbuffer = new unsigned char[1920 * 1080 * 3];
FILE* fp_source;
FILE* fp_rgb = NULL;
int l_nY, l_nU, l_nV;
double l_dR, l_dG, l_dB, l_ni;
fp_source = fopen("D:\\Sample_1920x1080.yuv", "rb");
// converting yuv file to rgb file
if (fp_source)
{
fp_rgb = fopen("D:\\Sample_1920x1080.rgb", "wb+");
while (!feof(fp_source))
{
fread(g_pcYUVBuffer, 1, l_nSize, fp_source);
unsigned char* l_pcY = g_pcYUVBuffer;
unsigned char* l_pcU = l_pcY + 1920 * 1080;
unsigned char* l_pcV = l_pcU + ((1920 * 1080) / 4);
unsigned char* l_pcRGBbuffer = g_pcRGBbuffer;
for (l_ni = 0; l_ni < (1920 * 1080) / 4; l_ni++)
{
l_nY = l_pcY[0];
l_nU = l_pcU[0];
l_nV = l_pcV[0];
l_dR = l_nY + 1.402 * (l_nV - 128);
l_dG = l_nY - 0.34414 * (l_nU - 128) - 0.71414 * (l_nV - 128);
l_dB = l_nY + 1.772 * (l_nU - 128);
// This prevents colour distortions in rgb image
if (l_dR < 0)
l_dR = 0;
else if (l_dR > 255)
l_dR = 255;
if (l_dG < 0)
l_dG = 0;
else if (l_dG > 255)
l_dG = 255;
if (l_dB < 0)
l_dB = 0;
else if (l_dB > 255)
l_dB = 255;
// 1st pixel of RGB
l_pcRGBbuffer[0] = l_dR;
l_pcRGBbuffer[1] = l_dG;
l_pcRGBbuffer[2] = l_dB;
l_nY = l_pcY[1];
l_nU = l_pcU[0];
l_nV = l_pcV[0];
l_dR = l_nY + 1.402 * (l_nV - 128);
l_dG = l_nY - 0.34414 * (l_nU - 128) - 0.71414 * (l_nV - 128);
l_dB = l_nY + 1.772 * (l_nU - 128);
if (l_dR < 0)
l_dR = 0;
else if (l_dR > 255)
l_dR = 255;
if (l_dG < 0)
l_dG = 0;
else if (l_dG > 255)
l_dG = 255;
if (l_dB < 0)
l_dB = 0;
else if (l_dB > 255)
l_dB = 255;
// 2nd pixel of RGB
l_pcRGBbuffer[3] = l_dR;
l_pcRGBbuffer[4] = l_dG;
l_pcRGBbuffer[5] = l_dB;
l_nY = l_pcY[2];
l_nU = l_pcU[0];
l_nV = l_pcV[0];
l_dR = l_nY + 1.402 * (l_nV - 128);
l_dG = l_nY - 0.34414 * (l_nU - 128) - 0.71414 * (l_nV - 128);
l_dB = l_nY + 1.772 * (l_nU - 128);
if (l_dR < 0)
l_dR = 0;
else if (l_dR > 255)
l_dR = 255;
if (l_dG < 0)
l_dG = 0;
else if (l_dG > 255)
l_dG = 255;
if (l_dB < 0)
l_dB = 0;
else if (l_dB > 255)
l_dB = 255;
// 3rd pixel of RGB
l_pcRGBbuffer[6] = l_dR;
l_pcRGBbuffer[7] = l_dG;
l_pcRGBbuffer[8] = l_dB;
l_nY = l_pcY[3];
l_nU = l_pcU[0];
l_nV = l_pcV[0];
// l_dR = 1.164*(l_nY-16 ) + 1.596*(l_nV-128 );
// l_dG = 1.164*(l_nY-16 ) - 0.813*(l_nV-128 ) - 0.391*(l_nU-128);
// l_dB = 1.164*(l_nY-16 ) + 2.018*(l_nU-128 );
l_dR = l_nY + 1.402 * (l_nV - 128);
l_dG = l_nY - 0.34414 * (l_nU - 128) - 0.71414 * (l_nV - 128);
l_dB = l_nY + 1.772 * (l_nU - 128);
if (l_dR < 0)
l_dR = 0;
else if (l_dR > 255)
l_dR = 255;
if (l_dG < 0)
l_dG = 0;
else if (l_dG > 255)
l_dG = 255;
if (l_dB < 0)
l_dB = 0;
else if (l_dB > 255)
l_dB = 255;
// 4th pixel of RGB
l_pcRGBbuffer[9] = l_dR;
l_pcRGBbuffer[10] = l_dG;
l_pcRGBbuffer[11] = l_dB;
l_pcY += 4;
l_pcU += 1;
l_pcV += 1;
l_pcRGBbuffer += 12;
}
fwrite(g_pcRGBbuffer, 1, 1920 * 1080 * 3, fp_rgb);
}
printf("Video converted to rgb file \n ");
}
else
{
printf("fail\n");
}
fclose(fp_rgb);
fclose(fp_source);
return 0;
}

Answer 1

Actually YUV 4:2:0 planar stores all Y pixels first,then U pixels,then V pixels
To extract correct pixels use the following forumla:
//Refer wikipedia for further details
size.total = size.width * size.height;
y = yuv[position.y * size.width + position.x];
u = yuv[(position.y / 2) * (size.width / 2) + (position.x / 2) + size.total];
v = yuv[(position.y / 2) * (size.width / 2) + (position.x / 2) + size.total + (size.total / 4)];
YUV2RGBTestApp2.cpp : Defines the entry point for the console application.
#include "stdafx.h"
#include <string>
#include <stdio.h>
unsigned char *g_pcRGBbuffer;
unsigned char *g_pcYUVBuffer;
int _tmain(int argc, _TCHAR* argv[])
{
int l_nSize = 1920 * 1080 * 1.5;
g_pcYUVBuffer = new unsigned char[l_nSize];
g_pcRGBbuffer = new unsigned char[1920 * 1080 * 3];
FILE *fp_source;
FILE *fp_rgb = NULL;
int l_ny, l_nu, l_nv, l_ni, RGBval;
int l_dr, l_dg, l_db;
fp_source = fopen("D:\\Sample_1920x1080.yuv", "rb");
int l_nj;
//converting yuv file to rgb file
if (fp_source) {
fp_rgb = fopen("D:\\Sample_1920x1080.rgb", "wb");
while (!feof(fp_source))
{
fread(g_pcYUVBuffer, 1, l_nSize, fp_source);
unsigned char *l_pcRGBbuffer = g_pcRGBbuffer;
for (int j = 0; j < 1080; j++)
{
for (int i = 0; i<1920; i++)
{
/*
Position for y,u,v components for yuv planar 4:2:0
Refer wikipedia for further reference
*/
int Y = g_pcYUVBuffer[j * 1920 + i];
int U = g_pcYUVBuffer[((j / 2) * 960) + (i / 2) + (1920 * 1080)];
int V = g_pcYUVBuffer[((j / 2) * 960) + (i / 2) + (1920 * 1080) + ((1920 * 1080) / 4)];
int R = 1.164*(Y - 16) + 1.596*(V - 128);
int G = 1.164*(Y - 16) - 0.813*(V - 128) - 0.391*(U - 128);
int B = 1.164*(Y - 16) + 2.018*(U - 128);
if (R>255)R = 255;
if (R<0)R = 0;
if (G>255)G = 255;
if (G<0)G = 0;
if (B>255)B = 255;
if (B<0)B = 0;
l_pcRGBbuffer[0] = R;
l_pcRGBbuffer[1] = G;
l_pcRGBbuffer[2] = B;
l_pcRGBbuffer += 3;
}
}
fwrite(g_pcRGBbuffer, 1, 1920 * 1080 * 3, fp_rgb);
}
printf("Video converted to rgb file \n ");
}
else {
printf("fail\n");
}
fclose(fp_rgb);
fclose(fp_source);
return 0;
}