Linux摄像头编程

这次是在linux下开发摄像头的程序,主要用的是video4linux来做的,界面用qt来实现,开始准备用framebuffer来直接写屏但是效果不怎么好,后来就用qt来做了,这样用起来效果还蛮好的,帧率也可以,可以上到30fps;运用v4l来编程主要掌握其api,要提高帧率最重要的是用到内存映射,其实用qt和framebuffer的时候都要用到内存映射来做,只有这样才可以达到较高的帧率,不过要注意资源的利用问题.mmap后一定要munmap.对于framebuffer是很有意思的一个东西,特别是驱动的设计.
代码如下:
1.open device:
video_dev = open("\dev\video0",O_RDWR));
2.get the information of the video device
struct video_capability video_cap;
memset(&video_cap,0,sizeof(video_cap));
if(ioctl(video_dev,VIDIOCGCAP,&video_cap) == -1)
{
perror("Cann't get the information of the video device");
close(video_dev);
exit(1);
}
3.get the information of the channels
struct video_channel video_chan;
memset(&video_chan,0,sizeof(video_chan));
for(channel = 0;channel < video_cap.channels;channel++)
{
video_chan.channel = channel;
if(ioctl(video_dev,VIDIOCGCHAN,&video_chan) == -1)
{
perror("Cann't get the information of the channels");
close(video_dev);
exit(3);
}
if(video_chan.type == VIDEO_TYPE_TV)
{
#ifdef DEBUG
printf("NO.%d channel is %s,type is
tv!\n",channel, video_chan.name);