1. 註冊全部容器格式和CODEC:av_register_all()
2. 打開文件:av_open_input_file()
3. 從文件中提取流信息:av_find_stream_info()
4. 窮舉全部的流,查找其中種類爲CODEC_TYPE_VIDEO
5. 查找對應的解碼器:avcodec_find_decoder()
6. 打開編解碼器:avcodec_open()
7. 爲解碼幀分配內存:avcodec_alloc_frame()
8. 不停地從碼流中提取出幀數據:av_read_frame()
9. 判斷幀的類型,對於視頻幀調用:avcodec_decode_video()
10. 解碼完後,釋放解碼器:avcodec_close()
11. 關閉輸入文件:av_close_input_file()android
FfmpegEncoder.hide
/* * FfmpegEncoder.h * * Current, Can Support YUV422sp encoder and decoder * * Created on: Dec 5, 2010 * Author: Henry.Wen */ #ifndef _H264ENCODER_H #define _H264ENCODER_H void save_image(const char* filePath, const void* bufferBase, int width, int height); int encoder_init(const char* filePath, int width, int height); int encoder_frame(const void* frame); int encoder_frame_yuv422(const void* frame); void encoder_close(); #endif
FfmpegEncoder.cpp函數
#include <math.h> #include <stdio.h> #include <stdlib.h> #include <string.h> extern "C" { #include <libavutil/mathematics.h> #include <libavformat/avformat.h> //#include <libswscale/swscale.h> } #include <skia/core/SkBitmap.h> #include <skia/images/SkImageEncoder.h> #include <android_runtime/AndroidRuntime.h> #include "FfmpegEncoder.h" AVOutputFormat *g_fmt = 0; AVFormatContext *g_oc = 0; AVCodec *g_video_codec = 0; AVStream *g_video_st = 0; AVFrame *g_frame = 0; AVPicture g_picture; int g_frame_count = 0; double g_video_pts = 0; int g_flagInit = 0; int g_width = 0; int g_height = 0; using namespace android; static Mutex sg_mutexLock; #ifndef LOGI #define LOGI(...) ((void)__android_log_print(ANDROID_LOG_INFO, "H264ENCODE", __VA_ARGS__)) #define LOGE(...) ((void)__android_log_print(ANDROID_LOG_ERROR, "H264ENCODE", __VA_ARGS__)) #endif void save_image(const char* filePath, const void* bufferBase, int width, int height) { Mutex::Autolock lock(sg_mutexLock); SkBitmap b; #if (ANDROID_r4_4_0) b.setConfig(SkBitmap::kARGB_8888_Config, width, height,(size_t)0); #else b.setConfig(SkBitmap::kARGB_8888_Config, width, height); #endif b.setPixels((void*)bufferBase); SkImageEncoder::EncodeFile(filePath, b, SkImageEncoder::kJPEG_Type, SkImageEncoder::kDefaultQuality); LOGI("save_image image ok===================="); } AVStream *add_stream(AVFormatContext *oc, AVCodec **codec, enum AVCodecID codec_id, int width, int height) { AVCodecContext *c; AVStream *st; /* find the encoder */ *codec = avcodec_find_encoder(codec_id); LOGI("encoder_init add_stream find encoder='%s'", avcodec_get_name(codec_id)); if (!(*codec)) { LOGE("encoder_init add_stream could not find encoder for '%s'\n", avcodec_get_name(codec_id)); return 0; } st = avformat_new_stream(oc, *codec); if (!st) { LOGE("encoder_init add_stream could not allocate stream"); return 0; } st->id = oc->nb_streams-1; c = st->codec; avcodec_get_context_defaults3(c, *codec); c->codec_id = codec_id;//AV_CODEC_ID_MPEG4; c->bit_rate = 40000; /* Resolution must be a multiple of two. */ c->width = width; c->height = height; c->time_base.den = 25; c->time_base.num = 1; c->gop_size = 12; /* emit one intra frame every twelve frames at most */ c->pix_fmt = AV_PIX_FMT_YUV420P; if (c->codec_id == AV_CODEC_ID_MPEG2VIDEO) { c->max_b_frames = 2; } if (c->codec_id == AV_CODEC_ID_MPEG1VIDEO) { c->mb_decision = 2; } /* Some formats want stream headers to be separate. */ if (oc->oformat->flags & AVFMT_GLOBALHEADER) c->flags |= CODEC_FLAG_GLOBAL_HEADER; if(!strcmp(oc->oformat->name, "mp4") || !strcmp(oc->oformat->name, "mov") || !strcmp(oc->oformat->name, "3gp")) c->flags |= CODEC_FLAG_GLOBAL_HEADER; return st; } AVFrame* alloc_picture(AVPixelFormat pix_fmt, int width, int height) { AVFrame *picture; uint8_t *picture_buf; int size; picture = avcodec_alloc_frame(); if (!picture) return NULL; size = avpicture_get_size(pix_fmt, width, height); picture_buf = (uint8_t *)av_malloc(size); if (!picture_buf) { av_free(picture); return NULL; } avpicture_fill((AVPicture *)picture, picture_buf, pix_fmt, width, height); return picture; } int open_video(AVFormatContext *oc, AVCodec *codec, AVStream *st) { int ret = 0; AVCodecContext *c = st->codec; /* open the codec */ ret = avcodec_open2(c, codec, NULL); if (ret < 0) { LOGE("encoder_init open_video could not open video codec: %s", av_err2str(ret)); return -1; } g_frame = avcodec_alloc_frame(); if (!g_frame) { LOGE("encoder_init open_video could not allocate video frame"); return -1; } ret = avpicture_alloc(&g_picture, c->pix_fmt, c->width, c->height); if (ret < 0) { LOGE("encoder_init open_video could not allocate picture: %s", av_err2str(ret)); free(g_frame); g_frame = 0; return -1; } *((AVPicture *)g_frame) = g_picture; return 0; } void close_video(AVStream *st) { if(st->codec) { avcodec_close(st->codec); st->codec = 0; } if(g_frame) { av_free(g_picture.data[0]); av_free(g_frame); g_frame = 0; } } int encoder_init(const char* filePath, int width, int height) { if(g_flagInit) return 0; Mutex::Autolock lock(sg_mutexLock); int ret = 0; LOGI("encoder_init ============begin"); if(!filePath || width <= 0 || height <= 0) { LOGE("encoder_init input parameters error ret = %d", (ret = -1)); return -1; } av_register_all(); /* allocate the output media context */ avformat_alloc_output_context2(&g_oc, NULL, NULL, filePath); if (!g_oc) { LOGI("Could not deduce output format from file extension: using MPEG."); return -1; } g_fmt = g_oc->oformat; g_video_st = add_stream(g_oc, &g_video_codec, g_fmt->video_codec, width, height); av_dump_format(g_oc, 0, filePath, 1); if (g_video_st) { if(open_video(g_oc, g_video_codec, g_video_st) < 0) { LOGE("encoder_init open_video fail!"); close_video(g_video_st); av_free(g_oc); g_oc = 0; return -1; } } else { LOGE("encoder_init g_video_st is null, not enough memory!"); av_free(g_oc); g_oc = 0; return -1; } /* open the output file, if needed */ if (!(g_fmt->flags & AVFMT_NOFILE)) { LOGI("encoder_init avio_open ============begin"); ret = avio_open(&g_oc->pb, filePath, AVIO_FLAG_WRITE); LOGI("encoder_init avio_open ret:%d============end", ret); if (ret < 0) { LOGE("encoder_init could not open '%s': %s", filePath, av_err2str(ret)); close_video(g_video_st); av_free(g_oc); g_oc = 0; return -1; } } LOGI("encoder_init avformat_write_header video file"); ret = avformat_write_header(g_oc, NULL); if (ret < 0) { LOGE("encoder_init error occurred when opening output file: %s\n", av_err2str(ret)); close_video(g_video_st); av_free(g_oc); g_oc = 0; return -1; } if (g_frame) g_frame->pts = 0; g_flagInit = 1; g_width = width; g_height = height; LOGI("encoder_init ============end"); return 0; } static struct SwsContext *swsContext; void fill_yuv_image(AVFrame *pict, const void* frame) { int x, y, tmpIndex = 0, tmpWdith = g_width >> 1, tmpHeight = g_height >> 1; unsigned char* tmpBuffer_yuv = (unsigned char*)frame; /* Y */ for (y = 0; y < g_height; ++y) { for (x = 0; x < g_width; ++x) { pict->data[0][y * pict->linesize[0] + x] = *(tmpBuffer_yuv + tmpIndex); ++tmpIndex; } } tmpIndex = 0; int tmpLength = g_width * g_height; unsigned char* tmpBuffer_uv = tmpBuffer_yuv + tmpLength; /* Cb and Cr */ for (y = 0; y < tmpHeight; ++y) { for (x = 0; x < tmpWdith; ++x) { pict->data[1][y * pict->linesize[1] + x] = *(tmpBuffer_uv + tmpIndex + 1); pict->data[2][y * pict->linesize[2] + x] = *(tmpBuffer_uv + tmpIndex); tmpIndex+= 2; } } } //fill AVFrame with YUV422p buffer void fill_yuv422p_image(AVFrame *pict, const void* frameYUV422p) { int width = g_width, height = g_height; unsigned char * pyuv422 = (unsigned char *)frameYUV422p; unsigned char * pyuv420y = &pict->data[0][0]; unsigned char * pyuv420u = &pict->data[1][0]; unsigned char * pyuv420v = &pict->data[2][0]; int uv_count = 0; int i, j; for (i = 0; i < height; i += 2) for (j = 0; j < width; j += 2) { memcpy(pyuv420y + i * width + j, pyuv422 + i * width * 2 + j * 2, 1); memcpy(pyuv420y + (i + 1) * width + j, pyuv422 + (i + 1) * width * 2 + j * 2, 1); memcpy(pyuv420y + i * width + (j + 1), pyuv422 + i * width * 2 + (j + 1) * 2, 1); memcpy(pyuv420y + (i + 1) * width + (j + 1), pyuv422 + (i + 1) * width * 2 + (j + 1) * 2, 1); //±£ÁôU ·ÖÁ¿ memcpy(pyuv420u + uv_count, pyuv422 + i * width * 2 + j * 2 + 1, 1); //±£ÁôV·ÖÁ¿; memcpy(pyuv420v + uv_count, pyuv422 + (i + 1) * width * 2 + (j + 1) * 2 + 1, 1); uv_count++; } /*int x, y, tmpIndex = 0, tmpWdith = g_width >> 1, tmpHeight = g_height >> 1; unsigned char* tmpBuffer_yuv = (unsigned char*)frameYUV422p; /* Y */ /*for (y = 0; y < g_height; ++y) { for (x = 0; x < g_width; ++x) { pict->data[0][y * pict->linesize[0] + x] = *(tmpBuffer_yuv + tmpIndex); ++tmpIndex; } } tmpIndex = 0; int tmpLength = g_width * g_height; unsigned char* tmpBuffer_uv = tmpBuffer_yuv + tmpLength; // Cb and Cr for (y = 0; y < tmpHeight; ++y) { for (x = 0; x < tmpWdith; ++x) { pict->data[1][y * pict->linesize[1] + x] = *(tmpBuffer_uv + tmpIndex); pict->data[2][y * pict->linesize[2] + x] = *(tmpBuffer_uv + tmpIndex +1); tmpIndex += 2; } tmpIndex += g_width; }*/ } void write_video_frame(AVFormatContext *oc, AVStream *st) { int ret; static struct SwsContext *sws_ctx; AVCodecContext *c = st->codec; /* encode the image */ AVPacket pkt; int got_output; av_init_packet(&pkt); pkt.data = NULL; // packet data will be allocated by the encoder pkt.size = 0; ret = avcodec_encode_video2(c, &pkt, g_frame, &got_output); if (ret < 0) { LOGE("encoder_init error encoding video frame: %s\n", av_err2str(ret)); return; } //If size is zero, it means the image was buffered. if (got_output) { if (c->coded_frame->key_frame) pkt.flags |= AV_PKT_FLAG_KEY; pkt.stream_index = st->index; ret = av_interleaved_write_frame(oc, &pkt); } else { ret = 0; } av_free_packet(&pkt); if (ret != 0) { LOGE("encoder_init error while writing video frame: %s\n", av_err2str(ret)); return; } ++g_frame_count = 0; } int encoder_frame(const void* frame) { if(!g_flagInit) return 0; Mutex::Autolock lock(sg_mutexLock); fill_yuv_image(g_frame, frame); if (g_video_st) g_video_pts = (double)g_video_st->pts.val * g_video_st->time_base.num / g_video_st->time_base.den; else g_video_pts = 0.0; write_video_frame(g_oc, g_video_st); g_frame->pts += av_rescale_q(1, g_video_st->codec->time_base, g_video_st->time_base); return 0; } int encoder_frame_yuv422(const void* frame) { if(!g_flagInit) return 0; Mutex::Autolock lock(sg_mutexLock); fill_yuv422p_image(g_frame, frame); if (g_video_st) g_video_pts = (double)g_video_st->pts.val * g_video_st->time_base.num / g_video_st->time_base.den; else g_video_pts = 0.0; write_video_frame(g_oc, g_video_st); g_frame->pts += av_rescale_q(1, g_video_st->codec->time_base, g_video_st->time_base); return 0; } void encoder_close() { LOGI("encoder_close ============begin"); Mutex::Autolock lock(sg_mutexLock); if(g_oc) { av_write_trailer(g_oc); if (g_video_st) close_video(g_video_st); for(int i = 0; i < (int)g_oc->nb_streams; ++i) { av_freep(&g_oc->streams[i]->codec); av_freep(&g_oc->streams[i]); } if (!(g_fmt->flags & AVFMT_NOFILE)) avio_close(g_oc->pb); av_free(g_oc); } g_oc = 0; g_video_st = 0; g_flagInit = 0; g_frame_count = 0; g_width = 0; g_height = 0; LOGI("encoder_close ============end"); }
首先第一件事情就是開一個視頻文件並從中獲得流。咱們要作的第一件事情就是使用av_register_all();來初始化libavformat/libavcodec: ui
這一步註冊庫中含有的全部可用的文件格式和編碼器,這樣當打開一個文件時,它們纔可以自動選擇相應的文件格式和編碼器。av_register_all()只需
調用一次,因此,要放在初始化代碼中。也能夠僅僅註冊我的的文件格式和編碼。
下一步,打開文件:
AVFormatContext *pFormatCtx;
const char *filename="myvideo.mpg";
av_open_input_file(&pFormatCtx, filename, NULL, 0, NULL); // 打開視頻文件
最後三個參數描述了文件格式,緩衝區大小(size)和格式參數;咱們經過簡單地指明NULL或0告訴 libavformat 去自動探測文件格式而且使用默認的緩
衝區大小。這裏的格式參數指的是視頻輸出參數,好比寬高的座標。
下一步,咱們須要取出包含在文件中的流信息:
av_find_stream_info(pFormatCtx); // 取出流信息
AVFormatContext 結構體
dump_format(pFormatCtx, 0, filename, false);//咱們可使用這個函數把獲取到得參數所有輸出。
for(i=0; i<pFormatCtx->nb_streams; i++) //區分視頻流和音頻流
if(pFormatCtx->streams->codec.codec_type==CODEC_TYPE_VIDEO) //找到視頻流,這裏也能夠換成音頻
{
videoStream=i;
break;
}
接下來就須要尋找解碼器
AVCodec *pCodec;
pCodec=avcodec_find_decoder(pCodecCtx->codec_id);
avcodec_open(pCodecCtx, pCodec); // 打開解碼器
給視頻幀分配空間以便存儲解碼後的圖片:
AVFrame *pFrame;
pFrame=avcodec_alloc_frame();
/////////////////////////////////////////開始解碼///////////////////////////////////////////
第一步固然是讀數據:
咱們將要作的是經過讀取包來讀取整個視頻流,而後把它解碼成幀,最後轉換格式而且保存。
while(av_read_frame(pFormatCtx, &packet)>=0) { //讀數據
if(packet.stream_index==videoStream){ //判斷是否視頻流
avcodec_decode_video(pCodecCtx,pFrame, &frameFinished,
packet.data, packet.size); //解碼
if(frameFinished) {
img_convert((AVPicture *)pFrameRGB, PIX_FMT_RGB24,(AVPicture*)pFrame, pCodecCtx->pix_fmt, pCodecCtx->width,pCodecCtx->height);//轉換
SaveFrame(pFrameRGB, pCodecCtx->width,pCodecCtx->height, i); //保存數據
av_free_packet(&packet); //釋放
av_read_frame()讀取一個包而且把它保存到AVPacket結構體中。這些數據能夠在後面經過av_free_packet()來釋 放。函數avcodec_decode_video()把包
轉換爲幀。然而當解碼一個包的時候,咱們可能沒有獲得咱們須要的關於幀的信息。所以,當咱們得 到下一幀的時候,avcodec_decode_video()爲咱們設
置了幀結束標誌frameFinished。最後,咱們使用 img_convert()函數來把幀從原始格式(pCodecCtx->pix_fmt)轉換成爲RGB格式。要記住,你能夠把一個
AVFrame結構體的指針轉換爲AVPicture結構體的指針。最後,咱們把幀和高度寬度信息傳遞給咱們的SaveFrame函數。
到此解碼完畢,顯示過程使用SDL完成考慮到咱們之後會使用firmware進行顯示操做,SDL忽略不講。
音視頻同步
DTS(解碼時間戳)和PTS(顯示時間戳)
當咱們調用av_read_frame()獲得一個包的時候,PTS和DTS的信息也會保存在包中。可是咱們真正想要的PTS是咱們剛剛解碼出來的 原始幀 的PTS,這樣我
們才能知道何時來顯示它。然而,咱們從avcodec_decode_video()函數中獲得的幀只是一個AVFrame,其中並 沒有包含有用的PTS值(注意:AVFrame並
沒有包含時間戳信息,但當咱們等到幀的時候並非咱們想要的樣子)。。咱們保存一幀的第一個包的PTS: 這將做爲整個這一幀的PTS。咱們 能夠經過函
數avcodec_decode_video()來計算出哪一個包是一幀的第一個包。怎樣實現呢?任什麼時候候當一個包開始一幀的時 候,avcodec_decode_video()將調用一個函數
來爲一幀申請一個緩衝。固然,ffmpeg容許咱們從新定義那個分配內存的函數。計算前 一幀和如今這一幀的時間戳來預測出下一個時間戳的時間。同時,我
們須要同步視頻到音頻。咱們將設置一個音頻時間audioclock;一個內部值記錄了我 們正在播放的音頻的位置。就像從任意的mp3播放器中讀出來的數字一
樣。既然咱們把視頻同步到音頻,視頻線程使用這個值來算出是否太快仍是太慢。編碼