c++ - 使用 FFMPEG 将每个输入图像转换为 H264 在 MevisLab 中运行的 Visual Studio 中编译时出错

Question

我正在 MevisLab 框架中创建一个 ML 模块，我使用 FFMPEG 将我获得的每个图像转换为 H264 视频，并在获得所有帧后保存它。但不幸的是，我在分配输出缓冲区大小时遇到​​问题。当我将其包含在代码中时，应用程序崩溃。如果我不包含它，则输出文件大小仅为 4kb。其中没有存储任何内容。

我也不太确定将 HBitmap 放入编码器是否是正确的方法。很高兴收到您的建议。

我的代码:

BITMAPINFO bitmapInfo;
            HDC        hdc;

            ZeroMemory(&bitmapInfo, sizeof(bitmapInfo));

            BITMAPINFOHEADER &bitmapInfoHeader = bitmapInfo.bmiHeader;
            bitmapInfoHeader.biSize            = sizeof(bitmapInfoHeader);
            bitmapInfoHeader.biWidth           = _imgWidth;
            bitmapInfoHeader.biHeight          = _imgHeight;
            bitmapInfoHeader.biPlanes          =  1;
            bitmapInfoHeader.biBitCount        = 24;
            bitmapInfoHeader.biCompression     = BI_RGB;
            bitmapInfoHeader.biSizeImage       = ((bitmapInfoHeader.biWidth * bitmapInfoHeader.biBitCount / 8 + 3) & 0xFFFFFFFC) * bitmapInfoHeader.biHeight;
            bitmapInfoHeader.biXPelsPerMeter   = 10000;
            bitmapInfoHeader.biYPelsPerMeter   = 10000;
            bitmapInfoHeader.biClrUsed         = 0;
            bitmapInfoHeader.biClrImportant    = 0;
            //RGBQUAD* Ref = new RGBQUAD[_imgWidth,_imgHeight];
            HDC hdcscreen = GetDC(0);

            hdc = CreateCompatibleDC(hdcscreen);
            ReleaseDC(0, hdcscreen);

            _hbitmap = CreateDIBSection(hdc, (BITMAPINFO*) &bitmapInfoHeader, DIB_RGB_COLORS, &_bits, NULL, NULL);

为了获取位图，我使用上面的代码。然后我按如下方式分配编解码器上下文

c->bit_rate = 400000;
                // resolution must be a multiple of two 
                c->width = 1920;
                c->height = 1080;
                // frames per second 
                frame_rate = _framesPerSecondFld->getIntValue();
                //AVRational rational = {1,10};
                //c->time_base = (AVRational){1,25};
                 //c->time_base = (AVRational){1,25};
                 c->gop_size = 10; // emit one intra frame every ten frames 
                 c->max_b_frames = 1;
                 c->keyint_min = 1;   //minimum GOP size
                 c->time_base.num = 1;                                  // framerate numerator
                 c->time_base.den = _framesPerSecondFld->getIntValue(); 
                 c->i_quant_factor = (float)0.71;                        // qscale factor between P and I frames
                 c->pix_fmt = AV_PIX_FMT_RGB32;
                 std::string msg;
                 msg.append("Context is stored");
                 _messageFld->setStringValue(msg.c_str());

我按照输入创建位图图像

PagedImage *inImg = getUpdatedInputImage(0);
        ML_CHECK(inImg);
        ImageVector imgExt = inImg->getImageExtent();
        if ((imgExt.x = _imgWidth) && (imgExt.y == _imgHeight))
        {
        if (((imgExt.x % 4)==0) && ((imgExt.y % 4) == 0))
        {
                 // read out input image and write output image into video
                // get input image as an array
                void* imgData = NULL;
                SubImageBox imageBox(imgExt); // get the whole image
                getTile(inImg, imageBox, MLuint8Type, &imgData);
                iData = (MLuint8*)imgData;
                int r = 0; int g = 0;int  b = 0;
                // since we have only images with
                // a z-ext of 1, we can compute the c stride as follows
                int cStride = _imgWidth * _imgHeight;
                uint8_t offset  = 0;
                // pointer into the bitmap that is
                // used to write images into the avi
                UCHAR* dst = (UCHAR*)_bits;
                for (int y = _imgHeight-1; y >= 0; y--)
                { // reversely scan the image. if y-rows of DIB are set in normal order, no compression will be available.
                    offset = _imgWidth * y;
                    for (int x = 0; x < _imgWidth; x++)
                    {
                        if (_isGreyValueImage)
                        {
                            r = iData[offset + x];
                            *dst++ = (UCHAR)r;
                            *dst++ = (UCHAR)r;
                            *dst++ = (UCHAR)r;
                        } 
                        else
                        {
                            b = iData[offset + x]; // windows bitmap need reverse order: bgr instead of rgb
                            g = iData[offset + x + cStride          ];
                            r = iData[offset + x + cStride + cStride];

                            *dst++ = (UCHAR)r;
                            *dst++ = (UCHAR)g;
                            *dst++ = (UCHAR)b;
                        }
                        // alpha channel in input image is ignored
                    }
                }

然后我将其添加到编码器中，如下所示写入 H264

 in_width   = c->width;
                 in_height  = c->height;
                 out_width  = c->width;
                 out_height = c->height;
                 ibytes = avpicture_get_size(PIX_FMT_BGR32, in_width, in_height);
                 obytes = avpicture_get_size(PIX_FMT_YUV420P, out_width, out_height);
                 outbuf_size = 100000 + c->width*c->height*(32>>3);      // allocate output buffer
                 outbuf = static_cast<uint8_t *>(malloc(outbuf_size));

                 if(!obytes)
                 {
                     std::string msg;
                     msg.append("Bytes cannot be allocated");
                     _messageFld->setStringValue(msg.c_str());
                 }
                 else
                 {
                     std::string msg;
                     msg.append("Bytes allocation done");
                     _messageFld->setStringValue(msg.c_str());
                 }
                 //create buffer for the output image
                 inbuffer  =  (uint8_t*)av_malloc(ibytes);
                 outbuffer =  (uint8_t*)av_malloc(obytes);
                 inbuffer  =  (uint8_t*)dst;

                 //create ffmpeg frame structures.  These do not allocate space for image data, 
                 //just the pointers and other information about the image.
                 AVFrame* inpic = avcodec_alloc_frame();
                 AVFrame* outpic = avcodec_alloc_frame();

                 //this will set the pointers in the frame structures to the right points in 
                 //the input and output buffers.
                 avpicture_fill((AVPicture*)inpic, inbuffer, PIX_FMT_BGR32, in_width, in_height);
                 avpicture_fill((AVPicture*)outpic, outbuffer, PIX_FMT_YUV420P, out_width, out_height);
                 av_image_alloc(outpic->data, outpic->linesize, c->width, c->height, c->pix_fmt, 1); 
                 inpic->data[0] += inpic->linesize[0]*(_imgHeight-1);                                                      // flipping frame
                 inpic->linesize[0] = -inpic->linesize[0];    

                 if(!inpic)
                 {
                     std::string msg;
                     msg.append("Image is empty");
                     _messageFld->setStringValue(msg.c_str());
                 }
                 else
                 {
                     std::string msg;
                     msg.append("Picture has allocations");
                     _messageFld->setStringValue(msg.c_str());
                 }

                 //create the conversion context
                 fooContext = sws_getContext(in_width, in_height, PIX_FMT_BGR32, out_width, out_height, PIX_FMT_YUV420P, SWS_FAST_BILINEAR, NULL, NULL, NULL);
                 //perform the conversion
                 sws_scale(fooContext, inpic->data, inpic->linesize, 0, in_height, outpic->data, outpic->linesize);
                 //out_size = avcodec_encode_video(c, outbuf,outbuf_size, outpic);
                 if(!out_size)
                 {
                     std::string msg;
                     msg.append("Outsize is not valid");
                     _messageFld->setStringValue(msg.c_str());
                 } 
                 else
                 {
                     std::string msg;
                     msg.append("Outsize is valid");
                     _messageFld->setStringValue(msg.c_str());
                 }
                     fwrite(outbuf, 1, out_size, f);
                     if(!fwrite)
                 {
                     std::string msg;
                     msg.append("Frames couldnt be written");
                     _messageFld->setStringValue(msg.c_str());
                 } 
                 else
                 {
                     std::string msg;
                     msg.append("Frames written to the file");
                     _messageFld->setStringValue(msg.c_str());
                 }
                    // for (;out_size; i++)
                    // {
                          out_size = avcodec_encode_video(c, outbuf, outbuf_size, NULL); 
                          std::string msg;                       
                          msg.append("Writing Frames");
                          _messageFld->setStringValue(msg.c_str());// encode the delayed frames
                          _numFramesFld->setIntValue(_numFramesFld->getIntValue()+1);
                          fwrite(outbuf, 1, out_size, f);
                    // }
                     outbuf[0] = 0x00;
                     outbuf[1] = 0x00;                                                                                               // add sequence end code to have a real mpeg file
                     outbuf[2] = 0x01;
                     outbuf[3] = 0xb7;
                     fwrite(outbuf, 1, 4, f);
}

然后关闭并清理图像缓冲区和文件

  ML_TRACE_IN("MovieCreator::_endRecording()")
if (_numFramesFld->getIntValue() == 0)
{
    _messageFld->setStringValue("Empty movie, nothing saved.");
} 
else 
{
    _messageFld->setStringValue("Movie written to disk.");
    _numFramesFld->setIntValue(0);
if (_hbitmap)
{ 
    DeleteObject(_hbitmap); 
}
if (c != NULL)
{
       av_free(outbuffer);     
       av_free(inpic);
       av_free(outpic);
       fclose(f);
       avcodec_close(c);                                                                                               // freeing memory
       free(outbuf);
       av_free(c);
}
}

}

我认为主要问题就在这里!!

                     //out_size = avcodec_encode_video(c, outbuf,outbuf_size, outpic);

Answer 1

H264 无法读取 RGB 帧。更改此行:

    c->pix_fmt = AV_PIX_FMT_RGB32

对此:

    c->pix_fmt = AV_PIX_FMT_YUV420P

还指定编码器的较少输入，让 FFMPEG 决定最佳设置。删除这些行并查看它是否有效:

    c->i_quant_factor = (float)0.71; 
    c->max_b_frames = 1;
    c->keyint_min = 1;

或者，您也可以尝试从这里的工作代码示例开始:

http://www.imc-store.com.au/Articles.asp?ID=276

该示例在 VS2010 中，使用 FFMPEG 将帧编码为 H264 编码的 AVI 文件。它有很多评论，我发现它非常有帮助。

您可以将从位图中读取的 BGR 字符数组传递到 FFMPEG 类中(只需将颜色交换为 RGB)。