如何將來自攝像機的輸入圖像編碼爲H.264流？

Question

我試圖使用Mac OS X 10.9.5上的libx264實時將來自MacBook Pro內置FaceTime高清攝像頭的輸入圖像實時編碼爲H.264視頻流。

下面是我所採取的步驟：在15fps的使用AVFoundation API（AVCaptureDevice類等）

使用libswscale將圖像轉換成看見320×180 YUV420P格式

1. 獲取1280×720從相機32BGRA圖像。
2. 使用libx264將圖像編碼爲H.264視頻流（基線配置文件）。

我每次從相機獲取圖像時都會應用上述步驟，相信編碼器會跟蹤編碼狀態並在可用時生成NAL單元。因爲我希望在向編碼器提供輸入圖像的同時獲得編碼幀，所以我決定每隔30幀（2秒）刷新一次編碼器（調用x264_encoder_delayed_frames（））以調用x264_encoder_delayed_frames（）。但是，當我重新啓動編碼時，編碼器會在一段時間後停止（x264_encoder_encode（）永不返回）。我嘗試在刷新之前更改幀數，但情況沒有改變。

下面是相關的代碼（我省略了圖像捕捉代碼，因爲它看起來沒有問題。）

你能指出什麼我可能是做錯了？

x264_t *encoder; 
x264_param_t param; 

// Will be called only first time. 
int initEncoder() { 
    int ret; 

    if ((ret = x264_param_default_preset(&param, "medium", NULL)) < 0) { 
    return ret; 
    } 

    param.i_csp = X264_CSP_I420; 
    param.i_width = 320; 
    param.i_height = 180; 
    param.b_vfr_input = 0; 
    param.b_repeat_headers = 1; 
    param.b_annexb = 1; 

    if ((ret = x264_param_apply_profile(&param, "baseline")) < 0) { 
    return ret; 
    } 

    encoder = x264_encoder_open(&param); 
    if (!encoder) { 
    return AVERROR_UNKNOWN; 
    } 

    return 0; 
} 

// Will be called from encodeFrame() defined below. 
int convertImage(const enum AVPixelFormat srcFmt, const int srcW, const int srcH, const uint8_t *srcData, const enum AVPixelFormat dstFmt, const int dstW, const int dstH, x264_image_t *dstData) { 
    struct SwsContext *sws_ctx; 
    int ret; 
    int src_linesize[4]; 
    uint8_t *src_data[4]; 

    sws_ctx = sws_getContext(srcW, srcH, srcFmt, 
         dstW, dstH, dstFmt, 
         SWS_BILINEAR, NULL, NULL, NULL); 

    if (!sws_ctx) { 
    return AVERROR_UNKNOWN; 
    } 

    if ((ret = av_image_fill_linesizes(src_linesize, srcFmt, srcW)) < 0) { 
    sws_freeContext(sws_ctx); 
    return ret; 
    } 

    if ((ret = av_image_fill_pointers(src_data, srcFmt, srcH, (uint8_t *) srcData, src_linesize)) < 0) { 
    sws_freeContext(sws_ctx); 
    return ret; 
    } 

    sws_scale(sws_ctx, (const uint8_t * const*)src_data, src_linesize, 0, srcH, dstData->plane, dstData->i_stride); 
    sws_freeContext(sws_ctx); 
    return 0; 
} 

// Will be called for each frame. 
int encodeFrame(const uint8_t *data, const int width, const int height) { 
    int ret; 
    x264_picture_t pic; 
    x264_picture_t pic_out; 
    x264_nal_t *nal; 
    int i_nal; 

    if ((ret = x264_picture_alloc(&pic, param.i_csp, param.i_width, param.i_height)) < 0) { 
    return ret; 
    } 

    if ((ret = convertImage(AV_PIX_FMT_RGB32, width, height, data, AV_PIX_FMT_YUV420P, 320, 180, &pic.img)) < 0) { 
    x264_picture_clean(&pic); 
    return ret; 
    } 

    if ((ret = x264_encoder_encode(encoder, &nal, &i_nal, &pic, &pic_out)) < 0) { 
    x264_picture_clean(&pic); 
    return ret; 
    } 

    if(ret) { 
    for (int i = 0; i < i_nal; i++) { 
     printNAL(nal + i); 
    } 
    } 

    x264_picture_clean(&pic); 
    return 0; 
} 

// Will be called every 30 frames. 
int flushEncoder() { 
    int ret; 
    x264_nal_t *nal; 
    int i_nal; 
    x264_picture_t pic_out; 

    /* Flush delayed frames */ 
    while (x264_encoder_delayed_frames(encoder)) { 
    if ((ret = x264_encoder_encode(encoder, &nal, &i_nal, NULL, &pic_out)) < 0) { 
     return ret; 
    } 

    if (ret) { 
     for (int j = 0; j < i_nal; j++) { 
     printNAL(nal + j); 
     } 
    } 
    } 
} 

Answer 1

當不存在更多輸入幀時，即在編碼結束時，您不應該在每幀之後刷新延遲幀，而只能刷新一次。