标准的V4L2 API
http://v4l.videotechnology.com/dwg/v4l2.pdf
在例程/home/dvevm_1_20/demos/ImageGray中,涉及到图像采集及显示的一些概念
主要的几个文件
capture.c
display.c
video.c
在demo里面采集用到的格式是UYVY
V4L2_PIX_FMT_UYVY ('UYVY')
V4L2_PIX_FMT_UYVY -- Variationof V4L2_PIX_FMT_YUYV with different order of samples in memory
In this format each four bytes is twopixels. Each four bytes is two Y's, a Cb and a Cr. Each Y goes to one of thepixels, and the Cb and Cr belong to both pixels. As you can see, the Cr and Cbcomponents have half the horizontal resolution of the Y component.
Example 2-1. V4L2_PIX_FMT_UYVY4 × 4 pixel image
Byte Order. Each cell is one byte.
Cb00 | Y'00 | Cr00 | Y'01 | Cb01 | Y'02 | Cr01 | Y'03 | |
start + 8: | Cb10 | Y'10 | Cr10 | Y'11 | Cb11 | Y'12 | Cr11 | Y'13 |
start + 16: | Cb20 | Y'20 | Cr20 | Y'21 | Cb21 | Y'22 | Cr21 | Y'23 |
start + 24: | Cb30 | Y'30 | Cr30 | Y'31 | Cb31 | Y'32 | Cr31 | Y'33 |
Color Sample Location.
0 |
| 1 |
| 2 |
| 3 | |
0 | Y | C | Y |
| Y | C | Y |
1 | Y | C | Y |
| Y | C | Y |
2 | Y | C | Y |
| Y | C | Y |
3 | Y | C | Y |
| Y | C | Y |
在imagegray里面把图片变成灰度是在filecopy_dec.c这个函数里面有这样的代码
static void PictureGray(void *pInbuf,void*pOutbuf,unsigned int len)
{
unsigned int i;
unsigned int * pIn = (unsigned int*)pInbuf;
unsigned int *pOut= (unsigned int*)pOutbuf;
len >>= 2;
for(i=0;i<len;i++)
{
*pOut= *pIn & 0xff00ff00 | 0x00800080;
pIn++;
pOut++;
}
}
从上面的UYVY的格式解释里面可以看到,变成灰度图像只是把图像四个字节跟0xff00ff00相与就可以了,一直没明白为什么还要在后面加上一个0x00800080相或,我原来一直以为是不是UYVY的格式解释不一样,找了半天各种格式之间区别的不少,讲到这个点子上的倒没看到,最后在TI的网站上有这样的回复
i want to transfer the input image ofyuv422 format to a gray image
if you want a grayscale output in the YCbCrcolorspace of your input you can just set the Cb and Cr values to 0x80and leave the Y as is, as Y is the brightness/luminance and is essentially thegreyscale version of the image so if you set the color values to a constantmedian value of 0x80 you end up with a grayscale image.
For working with the encodedecode demo, ifall you want to do is make the output look grayscale than the small functionmentioned above (and here) should still work for you. In the encodedecode demoyou are actually compressing and decompressing the video so you could performthis operation between capture and compress or between decompress and display,either way should work as the frame buffer will be in the proper YCbCR 4:2:2format at both points. In YCbCr 4:2:2 you have a stream of bytes in the formCbYCrYCbYCrYCbYCrY... and so on, where the Y values for luminance arefor each pixel and each 2 pixels shares a Cb and Cr for chrominance. Since agreyscale image is an image with only brightness/luminance and nochrominance/color all you have to do is remove the chrominance values,
howeversince the display is expecting an image formatted in color we cannot take themout completely,
but rather we can make the chrominancevalues constant, and for greyscale/B&W you want them to all be mid pointvalues of 0x80 (out of 0xFF).
The code below does just this, if yougive it a pointer to the YCbCr 4:2:2 frame buffer and the size of the buffer itwill step through and make all of the Cb and Cr values 0x80 so that the imageappears greyscale on the output.
Bernie Thompson:
//make the image greyscale by setting allchrominance to 0x80
void process_imagebw( void* currentFrame, int yRows, int xPixels)
{
int xx = 0;
for( xx = 0; xx < (yRows * xPixels)*2; xx+=2)//just operating on the chroma
{
*( ( (unsigned char*)currentFrame ) + xx ) = 0x80; //set all chromato midpoint 0x80
}
} // End process_imagebw()
对TVP5158采集到的YUV422分析,它是以UYVY格式存放的,因为U,V是正交化的由于本课题用到的图像处理基于黑白图像,所以只需对Y分量进行处理,那么第一步就是对YUV422提取出Y,如果要保留Y分量,那么需将UV置为0x80,如需置白/黑色,Y分量设为FF/00,提取出亮度信号后,即可作简单的二值化.
UV是色差分量,UV为0就会全是绿色,全为0x80的时候才能看到灰度图。