Кодирование живого видео в качестве H265 и потоковое его по RTP ⇐ C++

[Anonymous]

Я пытаюсь кодировать живое видео как H265 и отправить кодированный поток через RTP. Изображения вытянуты из камеры Basler Gige с использованием Pylon API. Они кодируются как H265 с использованием библиотеки X265, а затем отправляются через RTP с помощью библиотеки CCRTP. Я использую клиент GSTREAMER для тестирования его получения. Приложение/X-RTP, кодирование-name = h265! Fakesink Dimp = true в качестве клиента и получение этого после начала сервера:

Код: Выделить всё

gst-launch-1.0 -v udpsrc address=239.0.0.1 port=5004 ! application/x-rtp,encoding-name=H265 ! fakesink dump=true
Setting pipeline to PAUSED ...
Pipeline is live and does not need PREROLL ...
Pipeline is PREROLLED ...
Setting pipeline to PLAYING ...
New clock: GstSystemClock
/GstPipeline:pipeline0/GstCapsFilter:capsfilter0.GstPad:src: caps = application/x-rtp, encoding-name=(string)H265
/GstPipeline:pipeline0/GstFakeSink:fakesink0.GstPad:sink: caps = application/x-rtp, encoding-name=(string)H265
Redistribute latency...
00000000 (0x7fa31c0069b0): 80 60 04 c1 87 73 a4 fb a1 36 c7 9e 00 00 00 01  .`...s...6......
00000010 (0x7fa31c0069c0): 42 01 01 01 60 00 00 03 00 90 00 00 03 00 00 03  B...`...........
00000020 (0x7fa31c0069d0): 00 5a a0 05 a2 00 88 7d e5 ba 4a 4c 2e 01 00 00  .Z.....}..JL....
00000030 (0x7fa31c0069e0): 03 00 01 00 00 03 00 1e 08                       .........
00000000 (0x7fa31c019b00): 80 60 04 c2 87 75 04 8b a1 36 c7 9e 00 00 00 01  .`...u...6......
00000010 (0x7fa31c019b10): 44 01 c0 71 83 12                                D..q..
00000000 (0x7fa31c01a180): 80 60 04 c3 87 76 64 1b a1 36 c7 9e 00 00 01 4e  .`...vd..6.....N
00000010 (0x7fa31c01a190): 01 05 ff ff ff ff ff ff ff ff f8 2c a2 de 09 b5  ...........,....
00000020 (0x7fa31c01a1a0): 17 47 db bb 55 a4 fe 7f c2 fc 4e 78 32 36 35 20  .G..U.....Nx265
00000030 (0x7fa31c01a1b0): 28 62 75 69 6c 64 20 32 31 35 29 20 2d 20 34 2e  (build 215) - 4.
00000040 (0x7fa31c01a1c0): 31 2b 31 2d 31 64 31 31 37 62 65 3a 5b 4c 69 6e  1+1-1d117be:[Lin
00000050 (0x7fa31c01a1d0): 75 78 5d 5b 47 43 43 20 31 32 2e 32 2e 30 5d 5b  ux][GCC 12.2.0][
00000060 (0x7fa31c01a1e0): 36 34 20 62 69 74 5d 5b 6e 6f 61 73 6d 5d 20 38  64 bit][noasm] 8
00000070 (0x7fa31c01a1f0): 62 69 74 20 2d 20 48 2e 32 36 35 2f 48 45 56 43  bit - H.265/HEVC
00000080 (0x7fa31c01a200): 20 63 6f 64 65 63 20 2d 20 43 6f 70 79 72 69 67   codec - Copyrig
00000090 (0x7fa31c01a210): 68 74 20 32 30 31 33 2d 32 30 31 38 20 28 63 29  ht 2013-2018 (c)
000000a0 (0x7fa31c01a220): 20 4d 75 6c 74 69 63 6f 72 65 77 61 72 65 2c 20   Multicoreware,
000000b0 (0x7fa31c01a230): 49 6e 63 20 2d 20 68 74 74 70 3a 2f 2f 78 32 36  Inc - http://x26
000000c0 (0x7fa31c01a240): 35 2e 6f 72 67 20 2d 20 6f 70 74 69 6f 6e 73 3a  5.org - options:

etc.

Проблема заключается в том, что если я добавлю элемент rtph265depay к клиенту как GST-launch-1.0 -v UdpsRC Addrc = 239.0.0.1 Port = 5004! Приложение/X-RTP, кодирование-name = h265! rtph265depay! Fakesink Dimp = true Тогда это все, что я получаю после запуска сервера:

Код: Выделить всё

gst-launch-1.0 -v udpsrc address=239.0.0.1 port=5004 ! application/x-rtp,encoding-name=H265 ! rtph265depay ! fakesink dump=true
Setting pipeline to PAUSED ...
Pipeline is live and does not need PREROLL ...
Pipeline is PREROLLED ...
Setting pipeline to PLAYING ...
New clock: GstSystemClock
/GstPipeline:pipeline0/GstCapsFilter:capsfilter0.GstPad:src: caps = application/x-rtp, encoding-name=(string)H265, media=(string)video, clock-rate=(int)90000
/GstPipeline:pipeline0/GstRtpH265Depay:rtph265depay0.GstPad:sink: caps = application/x-rtp, encoding-name=(string)H265, media=(string)video, clock-rate=(int)90000
< /code>
Если я запускаю тестовый видео сервер, используя gst-launch-1.0 videoTestSrc! Видео/рентг! x265enc! H265Parse! rtph265pay config-interval = 1! udpsink host = 239.0.0.1 Port = 5004 

и используйте GST-Launch-1.0 -V адрес UDPSRC = 239.0.0.1 Port = 5004! Приложение/X-RTP, кодирование-name = h265! rtph265depay! Fakesink Dimp = true в качестве клиента, тогда это печатается:

Код: Выделить всё

gst-launch-1.0 -v udpsrc address=239.0.0.1 port=5004 ! application/x-rtp,encoding-name=H265 ! rtph265depay ! fakesink dump=true
Setting pipeline to PAUSED ...
Pipeline is live and does not need PREROLL ...
Pipeline is PREROLLED ...
Setting pipeline to PLAYING ...
New clock: GstSystemClock
/GstPipeline:pipeline0/GstCapsFilter:capsfilter0.GstPad:src: caps = application/x-rtp, encoding-name=(string)H265, media=(string)video, clock-rate=(int)90000
/GstPipeline:pipeline0/GstRtpH265Depay:rtph265depay0.GstPad:sink: caps = application/x-rtp, encoding-name=(string)H265, media=(string)video, clock-rate=(int)90000
/GstPipeline:pipeline0/GstRtpH265Depay:rtph265depay0.GstPad:src: caps = video/x-h265, stream-format=(string)hvc1, alignment=(string)au, codec_data=(buffer)01420101040800000300980800f000fcfff9f800000f03200001001740010c01ffff040800000300980800000300003f9598092100010030420101040800000300980800000300003f90014101e2294b2b3492657ff80008000b506060604000000300400000078222000100074401c172b46240
/GstPipeline:pipeline0/GstFakeSink:fakesink0.GstPad:sink: caps = video/x-h265, stream-format=(string)hvc1, alignment=(string)au, codec_data=(buffer)01420101040800000300980800f000fcfff9f800000f03200001001740010c01ffff040800000300980800000300003f9598092100010030420101040800000300980800000300003f90014101e2294b2b3492657ff80008000b506060604000000300400000078222000100074401c172b46240
Redistribute latency...
00000000 (0x7f7680023330): 00 00 09 0f 4e 01 05 ff ff ff ff ff ff ff ff ff  ....N...........
00000010 (0x7f7680023340): 0a 2c a2 de 09 b5 17 47 db bb 55 a4 fe 7f c2 fc  .,.....G..U.....
00000020 (0x7f7680023350): 4e 78 32 36 35 20 28 62 75 69 6c 64 20 31 39 39  Nx265 (build 199
00000030 (0x7f7680023360): 29 20 2d 20 33 2e 35 2b 31 2d 66 30 63 31 30 32  ) - 3.5+1-f0c102
00000040 (0x7f7680023370): 32 62 36 3a 5b 4c 69 6e 75 78 5d 5b 47 43 43 20  2b6:[Linux][GCC
00000050 (0x7f7680023380): 31 32 2e 32 2e 30 5d 5b 36 34 20 62 69 74 5d 20  12.2.0][64 bit]
00000060 (0x7f7680023390): 31 32 62 69 74 20 2d 20 48 2e 32 36 35 2f 48 45  12bit - H.265/HE
00000070 (0x7f76800233a0):  56 43 20 63 6f 64 65 63 20 2d 20 43 6f 70 79 72  VC codec - Copyr
00000080 (0x7f76800233b0): 69 67 68 74 20 32 30 31 33 2d 32 30 31 38 20 28  ight 2013-2018 (
00000090 (0x7f76800233c0): 63 29 20 4d 75 6c 74 69 63 6f 72 65 77 61 72 65  c) Multicoreware
000000a0 (0x7f76800233d0): 2c 20 49 6e 63 20 2d 20 68 74 74 70 3a 2f 2f 78  , Inc - http://x
000000b0 (0x7f76800233e0): 32 36 35 2e 6f 72 67 20 2d 20 6f 70 74 69 6f 6e  265.org - option
000000c0 (0x7f76800233f0): 73 3a 20 63 70 75 69 64 3d 31 31 31 31 30 33 39  s: cpuid=1111039

etc.
< /code>
Это код, который я пытаюсь отладить: < /p>
#include 
#include 

#include 

#include 
#include 

int main(int argc, char *argv[]) {
// Pylon initialization

Pylon::PylonAutoInitTerm autoInitTerm;

Pylon::CTlFactory& tlFactory = Pylon::CTlFactory::GetInstance();

Pylon::CDeviceInfo info;
//info.SetSerialNumber();
Pylon::DeviceInfoList_t filter;
filter.push_back(info);

Pylon::CBaslerUniversalInstantCamera camera;

int64_t gev_scps_packet_size = 1500;
int64_t gev_scpd = 50000;

// X265 initialization

printf("x265_max_bit_depth: %i\n", x265_max_bit_depth);
printf("x265_version_str: %s\n", x265_version_str);
printf("x265_build_info_str: %s\n", x265_build_info_str);

// RTP initialization

COMMONCPP_NAMESPACE::RTPSession* session = new COMMONCPP_NAMESPACE::RTPSession(COMMONCPP_NAMESPACE::InetHostAddress("0.0.0.0"));

session->setSchedulingTimeout(10000);
session->setExpireTimeout(1000000);
session->setPayloadFormat(COMMONCPP_NAMESPACE::DynamicPayloadFormat(96, 90000));

if (!session->addDestination(COMMONCPP_NAMESPACE::InetMcastAddress("239.0.0.1"), 5004)) {
printf("Could not connect to port.\n");
return -1;
}

session->startRunning();

if (session->isActive()) {
printf("The queue is active.\n");
}
else {
printf("The queue is not active.\n");
}

uint32 time_stamp = 0;

// Opens a file for testing the X265 encoding.

FILE *fp_dst = NULL;
fp_dst = fopen("tmp.h265", "wb");
if (fp_dst == NULL) {
return -1;
}

while (1) {
try {
Pylon::DeviceInfoList_t devices;

while (!(tlFactory.EnumerateDevices(devices, filter) >  0)) {
printf("The camera is not connected.\n");
Pylon::WaitObject::Sleep(5000);
}
printf("The camera is connected.\n");

printf("Creating device.\n");
Pylon::IPylonDevice* device_ptr = tlFactory.CreateFirstDevice(info);
if (device_ptr == nullptr) {
printf("tlFactory.CreateFirstDevice returned a null pointer.\n");
continue;
}

printf("Attaching device.\n");
camera.Attach(device_ptr);

printf("Opening camera.\n");
camera.Open();

int64_t width;
if (camera.Width.IsReadable()) {
width = camera.Width.GetValue();
printf("Width: %i\n", width);
}
else {
printf("Width is not readable.\n");
}

int64_t height;
if (camera.Height.IsReadable()) {
height = camera.Height.GetValue();
printf("Height: %i\n", height);
}
else {
printf("Height is not readable.\n");
}

Basler_UniversalCameraParams::PixelFormatEnums pixel_format;
if (camera.PixelFormat.IsReadable()) {
pixel_format = camera.PixelFormat.GetValue();
printf("Pixel Format: %s\n", camera.PixelFormat.ToString().c_str());
}
else {
printf("Pixel Format is not readable.\n");
}

Basler_UniversalCameraParams::PixelSizeEnums pixel_size;
if (camera.PixelSize.IsReadable()) {
pixel_size = camera.PixelSize.GetValue();
printf("Pixel Size: %s\n", camera.PixelSize.ToString().c_str());
}
else {
printf("Pixel Size is not readable.\n");
}

// Writes properties of the camera using the generic interface for cameras standard (GenICam).

if (camera.GevSCPSPacketSize.TrySetValue(gev_scps_packet_size)) {
printf("Set Gev SCPS Packet Size to %i.\n", gev_scps_packet_size);
}
else {
printf("Could not set Gev SCPS Packet Size.\n");
}

if (camera.GevSCPD.TrySetValue(gev_scpd)) {
printf("Set Gev SCPD (Inter-Packet Delay) to %i.\n", gev_scpd);
}
else {
printf("Could not set Gev SCPD (Inter-Packet Delay).\n");
}

// Configures the X265 encoder.

printf("Configuring the encoder.\n");

x265_param* param_ptr = x265_param_alloc();
if (param_ptr == nullptr) {
printf("x265_param_alloc returned a null pointer.\n");
continue;
}

x265_param_default(param_ptr);

if (x265_param_default_preset(param_ptr, "ultrafast", "zerolatency") < 0) {
printf("Error calling x265_param_default_preset.\n");
return -1;
}

/*
if (x265_param_apply_profile(param_ptr, "main") <  0) {
printf("Error calling x265_param_apply_profile.\n");
return -1;
}
*/

switch (pixel_size) {
case Basler_UniversalCameraParams::PixelSizeEnums::PixelSize_Bpp1 :
param_ptr->sourceBitDepth = 1;

case Basler_UniversalCameraParams::PixelSizeEnums::PixelSize_Bpp2 :
param_ptr->sourceBitDepth = 2;

case Basler_UniversalCameraParams::PixelSizeEnums::PixelSize_Bpp4 :
param_ptr->sourceBitDepth = 4;

case Basler_UniversalCameraParams::PixelSizeEnums::PixelSize_Bpp8 :
param_ptr->sourceBitDepth = 8;

case Basler_UniversalCameraParams::PixelSizeEnums::PixelSize_Bpp10 :
param_ptr->sourceBitDepth = 10;

case Basler_UniversalCameraParams::PixelSizeEnums::PixelSize_Bpp12 :
param_ptr->sourceBitDepth = 12;

case Basler_UniversalCameraParams::PixelSizeEnums::PixelSize_Bpp14 :
param_ptr->sourceBitDepth = 14;

case Basler_UniversalCameraParams::PixelSizeEnums::PixelSize_Bpp16 :
param_ptr->sourceBitDepth = 16;

case Basler_UniversalCameraParams::PixelSizeEnums::PixelSize_Bpp24 :
param_ptr->sourceBitDepth = 24;

case Basler_UniversalCameraParams::PixelSizeEnums::PixelSize_Bpp32 :
param_ptr->sourceBitDepth = 32;

case Basler_UniversalCameraParams::PixelSizeEnums::PixelSize_Bpp36 :
param_ptr->sourceBitDepth = 36;

case Basler_UniversalCameraParams::PixelSizeEnums::PixelSize_Bpp48 :
param_ptr->sourceBitDepth = 48;

case Basler_UniversalCameraParams::PixelSizeEnums::PixelSize_Bpp64 :
param_ptr->sourceBitDepth = 64;
}

param_ptr->sourceWidth = width;
param_ptr->sourceHeight = height;
param_ptr->fpsNum = 30;
param_ptr->fpsDenom = 1;
param_ptr->internalCsp = X265_CSP_I420;
param_ptr->bRepeatHeaders = true;

// param_ptr->keyframeMin = 25;
// param_ptr->keyframeMax = 250;

// param_ptr->scenecutThreshold = 40;
// param_ptr->bHistBasedSceneCut = 0;

// param_ptr->bEnableEndOfBitstream = false;
// param_ptr->bEnableEndOfSequence = false;

x265_picture* pic_in_ptr = x265_picture_alloc();
if (pic_in_ptr == nullptr) {
printf("x265_picture_alloc returned a null pointer.\n");
}
x265_picture_init(param_ptr, pic_in_ptr);

x265_picture* pic_out_ptr = x265_picture_alloc();
if (pic_out_ptr == nullptr) {
printf("x265_picture_alloc returned a null pointer.\n");
}

x265_encoder* encoder_ptr = x265_encoder_open(param_ptr);
if (encoder_ptr == nullptr) {
printf("x265_encoder_open returned a null pointer.\n");
}

x265_param_free(param_ptr);

x265_nal* pp_nal_ptr = nullptr;
uint32_t pi_nal = 0;

printf("Finished configuring the encoder.\n");

// Starts the loop for grabbing, encoding and sending images from the camera.

printf("Starting grabbing.\n");
camera.StartGrabbing(Pylon::GrabStrategy_LatestImageOnly);

Pylon::CBaslerUniversalGrabResultPtr ptrGrabResult;

while (1) {
try {
printf("Retrieving result.\n");
camera.RetrieveResult((unsigned int) 5000, ptrGrabResult, Pylon::TimeoutHandling_ThrowException);
}
catch (const Pylon::TimeoutException e) {
printf("Timeout exception calling RetrieveResult.\n");
printf("%s\n", e.GetDescription());

Pylon::WaitObject::Sleep(1000);

if (camera.IsCameraDeviceRemoved()) {
printf("The connection to the camera has been lost.\n");

camera.DestroyDevice();

while (x265_encoder_encode(encoder_ptr, &pp_nal_ptr, &pi_nal, 0, nullptr) >  0) {
printf("Flushing.\n");
}

x265_picture_free(pic_in_ptr);
x265_picture_free(pic_out_ptr);
x265_encoder_close(encoder_ptr);
x265_cleanup();

break;
}
}

if (ptrGrabResult->GrabSucceeded()) {
printf("Grab succeeded.\n");

Pylon::CImageFormatConverter image_format_converter;
image_format_converter.OutputPixelFormat = Pylon::PixelType_YUV420planar;
image_format_converter.OutputBitAlignment = Pylon::OutputBitAlignment_MsbAligned;

Pylon::CPylonImage image;

printf("Converting image.\n");
image_format_converter.Convert(
image,
ptrGrabResult->GetBuffer(),
ptrGrabResult->GetBufferSize(),
ptrGrabResult->GetPixelType(),
ptrGrabResult->GetWidth(),
ptrGrabResult->GetHeight(),
ptrGrabResult->GetPaddingX(),
Pylon::EImageOrientation::ImageOrientation_TopDown
);

size_t plane_0_stride;
Pylon::CPylonImage plane_0;
plane_0.CopyImage(image.GetPlane(0));
plane_0.GetStride(plane_0_stride);

size_t plane_1_stride;
Pylon::CPylonImage plane_1;
plane_1.CopyImage(image.GetPlane(1));
plane_1.GetStride(plane_1_stride);

size_t plane_2_stride;
Pylon::CPylonImage plane_2;
plane_2.CopyImage(image.GetPlane(2));
plane_2.GetStride(plane_2_stride);

pic_in_ptr->stride[0] = plane_0_stride;
pic_in_ptr->planes[0] = (void*) plane_0.GetBuffer();
pic_in_ptr->stride[1] = plane_1_stride;
pic_in_ptr->planes[1] = (void*) plane_1.GetBuffer();
pic_in_ptr->stride[2] = plane_2_stride;
pic_in_ptr->planes[2] = (void*) plane_2.GetBuffer();

printf("Encoding.\n");
int code = x265_encoder_encode(encoder_ptr, &pp_nal_ptr, &pi_nal, pic_in_ptr, pic_out_ptr);
if (code < 0) {
printf("Error encoding.\n");
return -1;
}
else if (code == 0) {
printf("The encoder pipeline is still filling or is empty after flushing.\n");
}
else {
for(int j = 0; j < pi_nal; j++) {
fwrite(pp_nal_ptr[j].payload, pp_nal_ptr[j].sizeBytes, 1, fp_dst);

printf("%u %u %u\n", j, pp_nal_ptr[j].type, pp_nal_ptr[j].sizeBytes);

session->putData(time_stamp, pp_nal_ptr[j].payload, pp_nal_ptr[j].sizeBytes);

// const char* buffer = "test";
// session->putData(time_stamp, (const unsigned char *) buffer, strlen((char*) buffer) + 1);

time_stamp += session->getCurrentRTPClockRate();
}
}
printf("Finished encoding.\n");
}
else {
printf("The grab failed.\n");
printf("%s\n", ptrGrabResult->GetErrorDescription().c_str());
}
}
}
catch (const Pylon::GenericException& e) {
printf("Generic exception.\n");
printf("%s\n", e.GetDescription());
}
}

return -1;
}

Я проверил, что сгенерированный тестовый файл может быть воспроизведен с помощью медиаплеера VLC и выглядит нормально.


Подробнее здесь: https://stackoverflow.com/questions/795 ... t-over-rtp