YUV420如何扣取指定区域的图像数据

YUV420数据的存储方式有两类：planar和packed。planar的YUV格式，先连续存储所有像素点的Y，紧接着存储所有像素点的U，随后是所有像素点的V。对于packed的YUV格式，每个像素点的Y,U,V是连续交*存储的。

本文记录一个对于planar的YUV420格式，大小为720P的RAW图像，读取指定区域的数据，并保存为一个新的YUV420格式RAW。

首先指定的新的宽高必须是4的整倍数，不然新图像会有底边会花，这是由YUV420这个数据存储格式决定的，如下图：

int CutBandYUV(uint8_t *pInBuf, uint32_t width, uint32_t height, uint8_t **pOutBuf, int new_x, int new_y, int new_width, int new_height)
{
if(NULL == pInBuf || 0 == width || 0 == height)
return -1;

int length = new_width* new_height* 3 / 2;
*pOutBuf = (uint8_t *)malloc(length);
if (NULL == *pOutBuf)
{
printf("malloc new size memory failed! size=%d\n", length);
return -1;
}
memset(*pOutBuf , 0, length);

uint8_t *pUBuf = *pOutBuf + new_width * new_height;
uint8_t *pVBuf = *pOutBuf +new_width * new_height * 5 / 4;
for(x = 0; x < new_width; x++)
{
for (y = 0; y < new_height; y++)	//每个循环写一列
{
*(*pOutBuf + y * new_width + x) = *(pInBuf + (x + new_x) + width * (y + new_y));	//cope Y
int ret = (y + new_y)%2;
if (1 == (x + new_x)%2 && 1 == (y + new_y)%2)
{
long pix = width * height + (width>>1) * ((y + new_y)>>1) + (((x + new_x))>>1);
*(pUBuf + (new_width/2)*(y/2) + x/2) = *(pInBuf + pix);	//cope U

pix += width * height / 4;
*(pVBuf + (new_width/2)*(y/2) + x/2) = *(pInBuf + pix);	//cope V
}
}
}
return 0;
}

本文出自 “大师兄” 博客，请务必保留此出处http://cto521.blog.51cto.com/9652841/1586190