ffmpeg應用例項：從拉流到推流（改進版）
阿新 • • 發佈：2019-02-16
說明：
1、使用 rtmp地址 和 rtsp地址 測試，程式可執行，生成的flv檔案可用VLC播放器播放；
2、音訊和視訊有些不同步；
3、網路推流不穩定，還有問題待解決；
#include "stdafx.h"
extern "C"
{
#include "libavcodec/avcodec.h"
#include "libavformat/avformat.h"
#include "libswscale/swscale.h"
#include "libswresample/swresample.h"
};
#include "opencv2/opencv.hpp"
#ifdef _DEBUG
#pragma comment(lib, "opencv_ts300d.lib")
#pragma comment(lib, "opencv_world300d.lib")
#else
#pragma comment(lib, "opencv_ts300.lib")
#pragma comment(lib, "opencv_world300.lib")
#endif
#pragma comment(lib, "avcodec.lib")
#pragma comment(lib, "avformat.lib")
#pragma comment(lib, "avutil.lib")
#pragma comment(lib, "swscale.lib")
#pragma comment(lib, "swresample.lib")
#define MAX_AUDIO_FRAME_SIZE 192000// 1 second of 48KHZ 32bit audio -> 32bit / 8 * 48000 * 1 = 192000 Byte
#define SAVE_FILE 0
/*
函式功能：將圖片 foreground 疊加到圖片 background 上
*/
static void overlayImage(const cv::Mat &background, const cv::Mat &foreground, cv::Mat &output, cv::Point2i location)
{
background.copyTo(output);
for (int y = std::max(location.y, 0); y < background.rows; ++y)
{
int fY = y - location.y;
if (fY >= foreground.rows)
break;
for (int x = std::max(location.x, 0); x < background.cols; ++x)
{
int fX = x - location.x;
if (fX >= foreground.cols)
break;
double opacity = ((double)foreground.data[fY * foreground.step + fX * foreground.channels() + 3]) / 255.;
for (int c = 0; opacity > 0 && c < output.channels(); ++c)
{
unsigned char foregroundPx =
foreground.data[fY * foreground.step + fX * foreground.channels() + c];
unsigned char backgroundPx =
background.data[y * background.step + x * background.channels() + c];
output.data[y*output.step + output.channels()*x + c] =
backgroundPx * (1. - opacity) + foregroundPx * opacity;
}
}
}
}
int main(int argc, char* argv[])
{
AVFormatContext* ifmt_ctx = NULL;
const char* input_filepath = NULL;
AVFormatContext* ofmt_ctx = NULL;
const char* output_filepath = NULL;
int idx_audio = -1;
int idx_video = -1;
AVStream* istream_audio = NULL;
AVCodecContext*codec_ctx_audio = NULL;
AVCodec* codec_audio = NULL;
AVStream* istream_video = NULL;
AVCodecContext*codec_ctx_video = NULL;
AVCodec* codec_video = NULL;
AVStream* ostream_video = NULL;
AVStream* ostream_audio = NULL;
AVCodecContext*codec_ctx_2h264 = NULL;
AVCodecContext*codec_ctx_2aac = NULL;
struct SwsContext* img_convert_ctx_2bgr = NULL;
struct SwsContext* img_convert_ctx_2yuv = NULL;
uint8_t* buffer_bgr = NULL;
uint8_t* buffer_yuv = NULL;
struct SwrContext* au_convert_ctx = NULL;
uint8_t* out_audio_buffer = NULL;
av_register_all();
avformat_network_init();
//input_filepath = "rtmp://live.hkstv.hk.lxdns.com/live/hks";
input_filepath = "rtsp://admin:
//output_filepath = "rtmp://127.0.0.1/live/mytest";
output_filepath = "output.flv";
AVDictionary* options = NULL;
av_dict_set(&options, "rtsp_transport", "tcp", 0);
if (avformat_open_input(&ifmt_ctx, input_filepath, NULL, &options) < 0)
{
printf("error: avformat_open_input\n");
goto END;
}
if (avformat_find_stream_info(ifmt_ctx, NULL) < 0)
{
printf("error: avformat_find_stream_info\n");
goto END;
}
if (avformat_alloc_output_context2(&ofmt_ctx, NULL, "flv", output_filepath) < 0)
{
printf("error: avformat_alloc_output_context2\n");
goto END;
}
for (int i = 0; i < ifmt_ctx->nb_streams; i++)
{
if (AVMEDIA_TYPE_AUDIO == ifmt_ctx->streams[i]->codec->codec_type)
{
idx_audio = i;
istream_audio = ifmt_ctx->streams[idx_audio];
codec_ctx_audio = istream_audio->codec;
ostream_audio = avformat_new_stream(ofmt_ctx, istream_audio->codec->codec);
avcodec_copy_context(ostream_audio->codec, istream_audio->codec);
ostream_audio->time_base = istream_audio->time_base;
ostream_audio->codec->codec_tag = 0;
if (ofmt_ctx->oformat->flags & AVFMT_GLOBALHEADER)
ostream_audio->codec->flags |= CODEC_FLAG_GLOBAL_HEADER;
}
else if (AVMEDIA_TYPE_VIDEO == ifmt_ctx->streams[i]->codec->codec_type)
{
idx_video = i;
istream_video = ifmt_ctx->streams[idx_video];
codec_ctx_video = istream_video->codec;
ostream_video = avformat_new_stream(ofmt_ctx, istream_video->codec->codec);
avcodec_copy_context(ostream_video->codec, istream_video->codec);
ostream_video->time_base = istream_video->time_base;
ostream_video->codec->codec_tag = 0;
if (ofmt_ctx->oformat->flags & AVFMT_GLOBALHEADER)
ostream_video->codec->flags |= CODEC_FLAG_GLOBAL_HEADER;
}
else
{
;
}
}
if (idx_video < 0 || idx_audio < 0)
{
printf("error: can not find any audio stream or video stream\n");
goto END;
}
codec_audio = avcodec_find_decoder(codec_ctx_audio->codec_id);
if (NULL == codec_audio)
{
printf("error: avcodec_find_decoder, input audio\n");
goto END;
}
codec_video = avcodec_find_decoder(codec_ctx_video->codec_id);
if (NULL == codec_video)
{
printf("error: avcodec_find_decoder, input video\n");
goto END;
}
if (avcodec_open2(codec_ctx_audio, codec_audio, NULL) < 0)// 開啟輸入音訊流的解碼器
{
printf("error: avcodec_open2, input audio\n");
goto END;
}
if (avcodec_open2(codec_ctx_video, codec_video, NULL) < 0)// 開啟輸入視訊流的解碼器
{
printf("error: avcodec_open2, input video\n");
goto END;
}
if (avio_open(&ofmt_ctx->pb, output_filepath, AVIO_FLAG_WRITE) < 0)
{
printf("error: avio_open\n");
goto END;
}
if (avformat_write_header(ofmt_ctx, NULL) < 0)
{
printf("error: avformat_write_header\n");
goto END;
}
av_dump_format(ifmt_ctx, -1, input_filepath, 0);
av_dump_format(ofmt_ctx, -1, output_filepath, 1);
AVFrame oframe_bgr = { 0 };
oframe_bgr.width = codec_ctx_video->width;
oframe_bgr.height = codec_ctx_video->height;
oframe_bgr.format = AV_PIX_FMT_BGR24;
int pic_size_bgr = avpicture_get_size((AVPixelFormat)oframe_bgr.format, oframe_bgr.width, oframe_bgr.height);
buffer_bgr = (uint8_t*)av_malloc(pic_size_bgr);
avpicture_fill((AVPicture*)&oframe_bgr, buffer_bgr, (AVPixelFormat)oframe_bgr.format, oframe_bgr.width, oframe_bgr.height);
img_convert_ctx_2bgr = sws_getContext(codec_ctx_video->width, codec_ctx_video->height, codec_ctx_video->pix_fmt,
oframe_bgr.width, oframe_bgr.height, (AVPixelFormat)oframe_bgr.format, SWS_BICUBIC, NULL, NULL, NULL);
AVFrame oframe_yuv = { 0 };
oframe_yuv.width = codec_ctx_video->width;
oframe_yuv.height = codec_ctx_video->height;
oframe_yuv.format = AV_PIX_FMT_YUV420P;
int pic_size_yuv = avpicture_get_size((AVPixelFormat)oframe_yuv.format, oframe_yuv.width, oframe_yuv.height);
buffer_yuv = (uint8_t*)av_malloc(pic_size_yuv);
avpicture_fill((AVPicture*)&oframe_yuv, buffer_yuv, (AVPixelFormat)oframe_yuv.format, oframe_yuv.width, oframe_yuv.height);
img_convert_ctx_2yuv = sws_getContext(oframe_bgr.width, oframe_bgr.height, (AVPixelFormat)oframe_bgr.format,
oframe_yuv.width, oframe_yuv.height, (AVPixelFormat)oframe_yuv.format, SWS_BICUBIC, NULL, NULL, NULL);
int64_t in_audio_channel_layout = av_get_default_channel_layout(codec_ctx_audio->channels);
AVSampleFormat in_audio_sample_fmt = codec_ctx_audio->sample_fmt;
int in_audio_sample_rate = codec_ctx_audio->sample_rate;
AVSampleFormat out_audio_sample_fmt = AV_SAMPLE_FMT_S16;
int out_audio_sample_rate = 44100;// <= 48000 HZ ?
int out_audio_nb_samples = codec_ctx_audio->frame_size;// AAC:1024 MP3:1152
uint64_t out_audio_channel_layout = AV_CH_LAYOUT_STEREO;
int out_audio_channels = av_get_channel_layout_nb_channels(out_audio_channel_layout);
int out_audio_buffer_size = av_samples_get_buffer_size(NULL, out_audio_channels, out_audio_nb_samples, out_audio_sample_fmt, 1);
out_audio_buffer = (uint8_t *)av_malloc(MAX_AUDIO_FRAME_SIZE * out_audio_channels);
au_convert_ctx = swr_alloc_set_opts(au_convert_ctx, out_audio_channel_layout, out_audio_sample_fmt, out_audio_sample_rate,
in_audio_channel_layout, in_audio_sample_fmt, in_audio_sample_rate, 0, NULL);
swr_init(au_convert_ctx);
AVCodec* codec_2h264 = NULL;
AVCodecID codec_id_2h264;
codec_id_2h264 = AV_CODEC_ID_H264;
codec_2h264 = avcodec_find_encoder(codec_id_2h264);
codec_ctx_2h264 = avcodec_alloc_context3(codec_2h264);
codec_ctx_2h264->codec_type = AVMEDIA_TYPE_VIDEO;
codec_ctx_2h264->pix_fmt = AV_PIX_FMT_YUV420P;
codec_ctx_2h264->bit_rate = codec_ctx_video->bit_rate;// 0 ？
codec_ctx_2h264->width = codec_ctx_video->width;
codec_ctx_2h264->height = codec_ctx_video->height;
codec_ctx_2h264->time_base = codec_ctx_video->time_base;
if (avcodec_open2(codec_ctx_2h264, codec_2h264, NULL) < 0)
{
printf("error: avcodec_open2, output video\n");
goto END;
}
AVCodec* codec_2aac = NULL;
AVCodecID codec_id_2aac;
codec_id_2aac = AV_CODEC_ID_AAC;
codec_2aac = avcodec_find_encoder(codec_id_2aac);
codec_ctx_2aac = avcodec_alloc_context3(codec_2aac);
codec_ctx_2aac->codec_type = AVMEDIA_TYPE_AUDIO;
codec_ctx_2aac->sample_fmt = AV_SAMPLE_FMT_FLTP;
codec_ctx_2aac->sample_rate = codec_ctx_audio->sample_rate;
codec_ctx_2aac->bit_rate = codec_ctx_audio->bit_rate;
codec_ctx_2aac->channel_layout = codec_ctx_audio->channel_layout;
codec_ctx_2aac->channels = codec_ctx_audio->channels;
codec_ctx_2aac->time_base = codec_ctx_audio->time_base;
if (avcodec_open2(codec_ctx_2aac, codec_2aac, NULL) < 0)
{
printf("error: avcodec_open2, output audio\n");
goto END;
}
AVPacket ipkt = { 0 };
av_init_packet(&ipkt);
ipkt.data = NULL;
ipkt.size = 0;
AVFrame iframe_audio = { 0 };
AVFrame iframe = { 0 };
AVPacket opkt_audio = { 0 };
av_init_packet(&opkt_audio);
opkt_audio.data = NULL;
opkt_audio.size = 0;
AVPacket opkt = { 0 };
av_init_packet(&opkt);
opkt.data = NULL;
opkt.size = 0;
#if SAVE_FILE
FILE* output_bgr = fopen("output.bgr", "wb+");
FILE* output_yuv = fopen("output.yuv", "wb+");
FILE* output_h264 = fopen("output.h264", "wb+");
FILE* output_pcm = fopen("output.pcm", "wb+");
FILE* output_aac = fopen("output.aac", "wb+");
#endif
while (1)
{
if (av_read_frame(ifmt_ctx, &ipkt) < 0)
break;
//if (0 == ipkt.pts)
//{
//continue;
//}
if (ipkt.stream_index == idx_audio)
{
avcodec_send_packet(codec_ctx_audio, &ipkt);
avcodec_receive_frame(codec_ctx_audio, &iframe_audio);
swr_convert(au_convert_ctx, &out_audio_buffer, MAX_AUDIO_FRAME_SIZE, (const uint8_t**)iframe_audio.data, iframe_audio.nb_samples);
#if SAVE_FILE
fwrite(out_audio_buffer, 1, out_audio_buffer_size, output_pcm);
#endif
iframe_audio.pts = ipkt.pts;
avcodec_send_frame(codec_ctx_2aac, &iframe_audio);
avcodec_receive_packet(codec_ctx_2aac, &opkt_audio);
#if SAVE_FILE
// 生成的AAC檔案無法播放。待解！
fwrite(opkt_audio.data, 1, opkt_audio.size, output_aac);
#endif
opkt_audio.pts = av_rescale_q_rnd(ipkt.pts, istream_audio->time_base, ostream_audio->time_base, (AVRounding)(AV_ROUND_NEAR_INF | AV_ROUND_PASS_MINMAX));
opkt_audio.dts = av_rescale_q_rnd(ipkt.dts, istream_audio->time_base, ostream_audio->time_base, (AVRounding)(AV_ROUND_NEAR_INF | AV_ROUND_PASS_MINMAX));
opkt_audio.duration = av_rescale_q(ipkt.duration, istream_audio->time_base, ostream_audio->time_base);
opkt_audio.pos = -1;
opkt_audio.stream_index = idx_audio;
av_interleaved_write_frame(ofmt_ctx, &opkt_audio);
av_free_packet(&opkt_audio);
static int audio_count = 0;
printf(" audio count = %d\n", ++audio_count);
}
else if (ipkt.stream_index == idx_video)
{
avcodec_send_packet(codec_ctx_video, &ipkt);
avcodec_receive_frame(codec_ctx_video, &iframe);
sws_scale(img_convert_ctx_2bgr, (const uint8_t* const*)iframe.data, iframe.linesize, 0, iframe.height, oframe_bgr.data, oframe_bgr.linesize);
cv::Mat background(oframe_bgr.height, oframe_bgr.width, CV_8UC3, oframe_bgr.data[0]);
overlayImage(background, cv::imread("logo.jpg"), background, cv::Point(20, 30));
#if SAVE_FILE
fwrite(oframe_bgr.data[0], (oframe_bgr.width) * (oframe_bgr.height) * 3, 1, output_bgr);
#endif
sws_scale(img_convert_ctx_2yuv, (const uint8_t* const*)oframe_bgr.data, oframe_bgr.linesize, 0, oframe_bgr.height, oframe_yuv.data, oframe_yuv.linesize);
#if SAVE_FILE
fwrite(oframe_yuv.data[0], (oframe_yuv.width) * (oframe_yuv.height), 1, output_yuv);
fwrite(oframe_yuv.data[1], (oframe_yuv.width) * (oframe_yuv.height) / 4, 1, output_yuv);
fwrite(oframe_yuv.data[2], (oframe_yuv.width) * (oframe_yuv.height) / 4, 1, output_yuv);
#endif
oframe_yuv.pts = ipkt.pts;
oframe_yuv.width = codec_ctx_2h264->width;
oframe_yuv.height = codec_ctx_2h264->height;
avcodec_send_frame(codec_ctx_2h264, &oframe_yuv);
avcodec_receive_packet(codec_ctx_2h264, &opkt);
#if SAVE_FILE
fwrite(opkt.data, 1, opkt.size, output_h264);
#endif
opkt.pts = av_rescale_q_rnd(ipkt.pts, istream_video->time_base, ostream_video->time_base, (AVRounding)(AV_ROUND_NEAR_INF | AV_ROUND_PASS_MINMAX));
opkt.dts = av_rescale_q_rnd(ipkt.dts, istream_video->time_base, ostream_video->time_base, (AVRounding)(AV_ROUND_NEAR_INF | AV_ROUND_PASS_MINMAX));
opkt.duration = av_rescale_q(ipkt.duration, istream_video->time_base, ostream_video->time_base);
opkt.pos = -1;
opkt.stream_index = idx_video;
av_interleaved_write_frame(ofmt_ctx, &opkt);
av_free_packet(&opkt);
static int video_count = 0;
printf("video_count = %d\n", ++video_count);
}
else
{
;
}
av_free_packet(&ipkt);
}
if (0 != av_write_trailer(ofmt_ctx))
{
printf("error: av_write_trailer\n");
goto END;
}
END:
#if FILE
if (NULL != output_pcm)
fclose(output_pcm);
if (NULL != output_aac)
fclose(output_aac);
if (NULL != output_bgr)
fclose(output_bgr);
if (NULL != output_yuv)
fclose(output_yuv);
if (NULL != output_h264)
fclose(output_h264);
#endif
avformat_close_input(&ifmt_ctx);
av_free(out_audio_buffer);
av_free(buffer_bgr);
av_free(buffer_yuv);
swr_free(&au_convert_ctx);
sws_freeContext(img_convert_ctx_2bgr);
sws_freeContext(img_convert_ctx_2yuv);
avcodec_free_context(&codec_ctx_2h264);
avcodec_free_context(&codec_ctx_2aac);
if (NULL != ofmt_ctx)
avio_close(ofmt_ctx->pb);
avformat_free_context(ofmt_ctx);
return 0;
}