Merge pull request #23596 from vrabaud:libavif

Add AVIF support through libavif. #23596

This is to fix https://github.com/opencv/opencv/issues/19271
Extra: https://github.com/opencv/opencv_extra/pull/1069

### Pull Request Readiness Checklist

See details at https://github.com/opencv/opencv/wiki/How_to_contribute#making-a-good-pull-request

- [x] I agree to contribute to the project under Apache 2 License.
- [x] To the best of my knowledge, the proposed patch is not based on a code under GPL or another license that is incompatible with OpenCV
- [x] The PR is proposed to the proper branch
- [x] There is a reference to the original bug report and related work
- [x] There is accuracy test, performance test and test data in opencv_extra repository, if applicable
      Patch to opencv_extra has the same branch name.
- [x] The feature is well documented and sample code can be built with the project CMake

CMakeLists.txt

@@ -240,6 +240,8 @@ OCV_OPTION(WITH_1394 "Include IEEE1394 support" ON
 OCV_OPTION(WITH_AVFOUNDATION "Use AVFoundation for Video I/O (iOS/Mac)" ON
   VISIBLE_IF APPLE
   VERIFY HAVE_AVFOUNDATION)
+OCV_OPTION(WITH_AVIF "Enable AVIF support" OFF
+  VERIFY HAVE_AVIF)
 OCV_OPTION(WITH_CAP_IOS "Enable iOS video capture" ON
   VISIBLE_IF IOS
   VERIFY HAVE_CAP_IOS)
@@ -1379,6 +1381,14 @@ if(WITH_WEBP OR HAVE_WEBP)
   status("    WEBP:" WEBP_FOUND THEN "${WEBP_LIBRARY} (ver ${WEBP_VERSION})" ELSE "build (ver ${WEBP_VERSION})")
 endif()
 
+if(WITH_AVIF OR HAVE_AVIF)
+  if(AVIF_VERSION)
+    status("    AVIF:" AVIF_FOUND THEN "${AVIF_LIBRARY} (ver ${AVIF_VERSION})" ELSE "NO")
+  else()
+    status("    AVIF:" AVIF_FOUND THEN "${AVIF_LIBRARY}" ELSE "NO")
+  endif()
+endif()
+
 if(WITH_PNG OR HAVE_PNG OR WITH_SPNG)
   if(WITH_SPNG)
     status("    PNG:" "build-${SPNG_LIBRARY} (ver ${SPNG_VERSION})")

cmake/OpenCVFindAVIF.cmake

@@ -0,0 +1,46 @@
+#=============================================================================
+# Find AVIF library
+#=============================================================================
+# Find the native AVIF headers and libraries.
+#
+#  AVIF_INCLUDE_DIRS - where to find avif/avif.h, etc.
+#  AVIF_LIBRARIES    - List of libraries when using AVIF.
+#  AVIF_FOUND        - True if AVIF is found.
+#=============================================================================
+
+# Look for the header file.
+
+unset(AVIF_FOUND)
+
+find_package(libavif QUIET)
+
+if(TARGET avif)
+    MARK_AS_ADVANCED(AVIF_INCLUDE_DIR)
+    MARK_AS_ADVANCED(AVIF_LIBRARY)
+
+    SET(AVIF_FOUND TRUE)
+    GET_TARGET_PROPERTY(AVIF_LIBRARY avif LOCATION)
+    GET_TARGET_PROPERTY(AVIF_INCLUDE_DIR1 avif INCLUDE_DIRECTORIES)
+    GET_TARGET_PROPERTY(AVIF_INCLUDE_DIR2 avif INTERFACE_INCLUDE_DIRECTORIES)
+    set(AVIF_INCLUDE_DIR)
+    if(AVIF_INCLUDE_DIR1)
+        LIST(APPEND AVIF_INCLUDE_DIR ${AVIF_INCLUDE_DIR1})
+    endif()
+    if(AVIF_INCLUDE_DIR2)
+        LIST(APPEND AVIF_INCLUDE_DIR ${AVIF_INCLUDE_DIR2})
+    endif()
+else()
+    FIND_PATH(AVIF_INCLUDE_DIR NAMES avif/avif.h)
+
+     # Look for the library.
+    FIND_LIBRARY(AVIF_LIBRARY NAMES avif)
+    MARK_AS_ADVANCED(AVIF_LIBRARY)
+
+    # handle the QUIETLY and REQUIRED arguments and set AVIF_FOUND to TRUE if
+    # all listed variables are TRUE
+    INCLUDE(${CMAKE_ROOT}/Modules/FindPackageHandleStandardArgs.cmake)
+    FIND_PACKAGE_HANDLE_STANDARD_ARGS(AVIF DEFAULT_MSG AVIF_LIBRARY AVIF_INCLUDE_DIR)
+
+    SET(AVIF_LIBRARIES ${AVIF_LIBRARY})
+    SET(AVIF_INCLUDE_DIRS ${AVIF_INCLUDE_DIR})
+endif()

cmake/OpenCVFindLibsGrfmt.cmake

@@ -37,6 +37,16 @@ if(NOT ZLIB_FOUND)
   ocv_parse_header2(ZLIB "${${ZLIB_LIBRARY}_SOURCE_DIR}/zlib.h" ZLIB_VERSION)
 endif()
 
+# --- libavif (optional) ---
+
+if(WITH_AVIF)
+  ocv_clear_internal_cache_vars(AVIF_LIBRARY AVIF_INCLUDE_DIR)
+  include(cmake/OpenCVFindAVIF.cmake)
+  if(AVIF_FOUND)
+    set(HAVE_AVIF 1)
+  endif()
+endif()
+
 # --- libjpeg (optional) ---
 if(WITH_JPEG)
   if(BUILD_JPEG)

cmake/templates/cvconfig.h.in

@@ -72,6 +72,9 @@
 #cmakedefine HAVE_OPENJPEG
 #cmakedefine HAVE_JASPER
 
+/* AVIF codec */
+#cmakedefine HAVE_AVIF
+
 /* IJG JPEG codec */
 #cmakedefine HAVE_JPEG
 

modules/imgcodecs/CMakeLists.txt

@@ -13,6 +13,11 @@ if(HAVE_WINRT_CX AND NOT WINRT)
   set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} /ZW")
 endif()
 
+if (HAVE_AVIF)
+  ocv_include_directories(${AVIF_INCLUDE_DIR})
+  list(APPEND GRFMT_LIBS ${AVIF_LIBRARY})
+endif()
+
 if(HAVE_JPEG)
   ocv_include_directories(${JPEG_INCLUDE_DIR} ${${JPEG_LIBRARY}_BINARY_DIR})
   list(APPEND GRFMT_LIBS ${JPEG_LIBRARIES})

modules/imgcodecs/include/opencv2/imgcodecs.hpp

@@ -105,7 +105,10 @@ enum ImwriteFlags {
        IMWRITE_TIFF_XDPI           = 257,//!< For TIFF, use to specify the X direction DPI
        IMWRITE_TIFF_YDPI           = 258,//!< For TIFF, use to specify the Y direction DPI
        IMWRITE_TIFF_COMPRESSION    = 259,//!< For TIFF, use to specify the image compression scheme. See libtiff for integer constants corresponding to compression formats. Note, for images whose depth is CV_32F, only libtiff's SGILOG compression scheme is used. For other supported depths, the compression scheme can be specified by this flag; LZW compression is the default.
-       IMWRITE_JPEG2000_COMPRESSION_X1000 = 272 //!< For JPEG2000, use to specify the target compression rate (multiplied by 1000). The value can be from 0 to 1000. Default is 1000.
+       IMWRITE_JPEG2000_COMPRESSION_X1000 = 272,//!< For JPEG2000, use to specify the target compression rate (multiplied by 1000). The value can be from 0 to 1000. Default is 1000.
+       IMWRITE_AVIF_QUALITY        = 512,//!< For AVIF, it can be a quality between 0 and 100 (the higher the better). Default is 95.
+       IMWRITE_AVIF_DEPTH          = 513,//!< For AVIF, it can be 8, 10 or 12. If >8, it is stored/read as CV_32F. Default is 8.
+       IMWRITE_AVIF_SPEED          = 514 //!< For AVIF, it is between 0 (slowest) and (fastest). Default is 9.
      };
 
 enum ImwriteJPEGSamplingFactorParams {
@@ -185,6 +188,7 @@ Currently, the following file formats are supported:
 -   JPEG 2000 files - \*.jp2 (see the *Note* section)
 -   Portable Network Graphics - \*.png (see the *Note* section)
 -   WebP - \*.webp (see the *Note* section)
+-   AVIF - \*.avif (see the *Note* section)
 -   Portable image format - \*.pbm, \*.pgm, \*.ppm \*.pxm, \*.pnm (always supported)
 -   PFM files - \*.pfm (see the *Note* section)
 -   Sun rasters - \*.sr, \*.ras (always supported)

modules/imgcodecs/src/grfmt_avif.cpp

@@ -0,0 +1,369 @@
+// This file is part of OpenCV project.
+// It is subject to the license terms in the LICENSE file found in the top-level
+// directory of this distribution and at http://opencv.org/license.html
+
+#include "precomp.hpp"
+
+#ifdef HAVE_AVIF
+
+#include <avif/avif.h>
+#include <fstream>
+
+#include <opencv2/core/utils/configuration.private.hpp>
+#include "opencv2/imgproc.hpp"
+#include "grfmt_avif.hpp"
+
+#define CV_AVIF_USE_QUALITY \
+  (AVIF_VERSION > ((0 * 1000000) + (11 * 10000) + (1 * 100)))
+
+#if !CV_AVIF_USE_QUALITY
+#define AVIF_QUALITY_LOSSLESS 100
+#define AVIF_QUALITY_WORST 0
+#define AVIF_QUALITY_BEST 100
+
+#endif
+
+namespace cv {
+namespace {
+
+struct AvifImageDeleter {
+  void operator()(avifImage *image) { avifImageDestroy(image); }
+};
+
+using AvifImageUniquePtr = std::unique_ptr<avifImage, AvifImageDeleter>;
+
+avifResult CopyToMat(const avifImage *image, int channels, Mat *mat) {
+  CV_Assert((int)image->height == mat->rows);
+  CV_Assert((int)image->width == mat->cols);
+  if (channels == 1) {
+    const cv::Mat image_wrap =
+        cv::Mat(image->height, image->width,
+                CV_MAKE_TYPE((image->depth == 8) ? CV_8U : CV_16U, 1),
+                image->yuvPlanes[0], image->yuvRowBytes[0]);
+    if ((image->depth == 8 && mat->depth() == CV_8U) ||
+        (image->depth > 8 && mat->depth() == CV_16U)) {
+      image_wrap.copyTo(*mat);
+    } else {
+      CV_Assert(image->depth > 8 && mat->depth() == CV_8U);
+      image_wrap.convertTo(*mat, CV_8U, 1. / (1 << (image->depth - 8)));
+    }
+    return AVIF_RESULT_OK;
+  }
+  avifRGBImage rgba;
+  avifRGBImageSetDefaults(&rgba, image);
+  if (channels == 3) {
+    rgba.format = AVIF_RGB_FORMAT_BGR;
+  } else {
+    CV_Assert(channels == 4);
+    rgba.format = AVIF_RGB_FORMAT_BGRA;
+  }
+  rgba.rowBytes = mat->step[0];
+  rgba.depth = (mat->depth() == CV_16U) ? image->depth : 8;
+  rgba.pixels = reinterpret_cast<uint8_t *>(mat->data);
+  return avifImageYUVToRGB(image, &rgba);
+}
+
+AvifImageUniquePtr ConvertToAvif(const cv::Mat &img, bool lossless,
+                                 int bit_depth) {
+  CV_Assert(img.depth() == CV_8U || img.depth() == CV_16U);
+
+  const int width = img.cols;
+  const int height = img.rows;
+
+  avifImage *result;
+
+  if (img.channels() == 1) {
+    result = avifImageCreateEmpty();
+    if (result == nullptr) return nullptr;
+    result->width = width;
+    result->height = height;
+    result->depth = bit_depth;
+    result->yuvFormat = AVIF_PIXEL_FORMAT_YUV400;
+    result->colorPrimaries = AVIF_COLOR_PRIMARIES_UNSPECIFIED;
+    result->transferCharacteristics = AVIF_TRANSFER_CHARACTERISTICS_UNSPECIFIED;
+    result->matrixCoefficients = AVIF_MATRIX_COEFFICIENTS_IDENTITY;
+    result->yuvRange = AVIF_RANGE_FULL;
+    result->yuvPlanes[0] = img.data;
+    result->yuvRowBytes[0] = img.step[0];
+    result->imageOwnsYUVPlanes = AVIF_FALSE;
+    return AvifImageUniquePtr(result);
+  }
+
+  if (lossless) {
+    result =
+        avifImageCreate(width, height, bit_depth, AVIF_PIXEL_FORMAT_YUV444);
+    if (result == nullptr) return nullptr;
+    result->colorPrimaries = AVIF_COLOR_PRIMARIES_UNSPECIFIED;
+    result->transferCharacteristics = AVIF_TRANSFER_CHARACTERISTICS_UNSPECIFIED;
+    result->matrixCoefficients = AVIF_MATRIX_COEFFICIENTS_IDENTITY;
+    result->yuvRange = AVIF_RANGE_FULL;
+  } else {
+    result =
+        avifImageCreate(width, height, bit_depth, AVIF_PIXEL_FORMAT_YUV420);
+    if (result == nullptr) return nullptr;
+    result->colorPrimaries = AVIF_COLOR_PRIMARIES_BT709;
+    result->transferCharacteristics = AVIF_TRANSFER_CHARACTERISTICS_SRGB;
+    result->matrixCoefficients = AVIF_MATRIX_COEFFICIENTS_BT601;
+    result->yuvRange = AVIF_RANGE_FULL;
+  }
+
+  avifRGBImage rgba;
+  avifRGBImageSetDefaults(&rgba, result);
+  if (img.channels() == 3) {
+    rgba.format = AVIF_RGB_FORMAT_BGR;
+  } else {
+    CV_Assert(img.channels() == 4);
+    rgba.format = AVIF_RGB_FORMAT_BGRA;
+  }
+  rgba.rowBytes = img.step[0];
+  rgba.depth = bit_depth;
+  rgba.pixels =
+      const_cast<uint8_t *>(reinterpret_cast<const uint8_t *>(img.data));
+
+  if (avifImageRGBToYUV(result, &rgba) != AVIF_RESULT_OK) {
+    avifImageDestroy(result);
+    return nullptr;
+  }
+  return AvifImageUniquePtr(result);
+}
+
+}  // namespace
+
+// 64Mb limit to avoid memory saturation.
+static const size_t kParamMaxFileSize = utils::getConfigurationParameterSizeT(
+    "OPENCV_IMGCODECS_AVIF_MAX_FILE_SIZE", 64 * 1024 * 1024);
+
+static constexpr size_t kAvifSignatureSize = 500;
+
+AvifDecoder::AvifDecoder() {
+  m_buf_supported = true;
+  channels_ = 0;
+  decoder_ = avifDecoderCreate();
+}
+
+AvifDecoder::~AvifDecoder() {
+  if (decoder_ != nullptr) avifDecoderDestroy(decoder_);
+}
+
+size_t AvifDecoder::signatureLength() const { return kAvifSignatureSize; }
+
+bool AvifDecoder::checkSignature(const String &signature) const {
+  avifDecoderSetIOMemory(decoder_,
+                         reinterpret_cast<const uint8_t *>(signature.c_str()),
+                         signature.size());
+  decoder_->io->sizeHint = 1e9;
+  const avifResult status = avifDecoderParse(decoder_);
+  return (status == AVIF_RESULT_OK || status == AVIF_RESULT_TRUNCATED_DATA);
+}
+
+#define OPENCV_AVIF_CHECK_STATUS(X, ENCDEC)               \
+  {                                                       \
+    const avifResult status = (X);                        \
+    if (status != AVIF_RESULT_OK) {                       \
+      const std::string error(ENCDEC->diag.error);        \
+      CV_Error(Error::StsParseError,                      \
+               error + " " + avifResultToString(status)); \
+      return false;                                       \
+    }                                                     \
+  }
+
+ImageDecoder AvifDecoder::newDecoder() const { return makePtr<AvifDecoder>(); }
+
+bool AvifDecoder::readHeader() {
+  if (!m_buf.empty()) {
+    CV_Assert(m_buf.type() == CV_8UC1);
+    CV_Assert(m_buf.rows == 1);
+  }
+
+  OPENCV_AVIF_CHECK_STATUS(
+      m_buf.empty()
+          ? avifDecoderSetIOFile(decoder_, m_filename.c_str())
+          : avifDecoderSetIOMemory(
+                decoder_, reinterpret_cast<const uint8_t *>(m_buf.data),
+                m_buf.total()),
+      decoder_);
+  OPENCV_AVIF_CHECK_STATUS(avifDecoderParse(decoder_), decoder_);
+
+  m_width = decoder_->image->width;
+  m_height = decoder_->image->height;
+  channels_ = (decoder_->image->yuvFormat == AVIF_PIXEL_FORMAT_YUV400) ? 1 : 3;
+  if (decoder_->alphaPresent) ++channels_;
+  bit_depth_ = decoder_->image->depth;
+  CV_Assert(bit_depth_ == 8 || bit_depth_ == 10 || bit_depth_ == 12);
+  m_type = CV_MAKETYPE(bit_depth_ == 8 ? CV_8U : CV_16U, channels_);
+  is_first_image_ = true;
+  return true;
+}
+
+bool AvifDecoder::readData(Mat &img) {
+  CV_CheckGE(m_width, 0, "");
+  CV_CheckGE(m_height, 0, "");
+
+  CV_CheckEQ(img.cols, m_width, "");
+  CV_CheckEQ(img.rows, m_height, "");
+  CV_CheckType(
+      img.type(),
+      (img.channels() == 1 || img.channels() == 3 || img.channels() == 4) &&
+          (img.depth() == CV_8U || img.depth() == CV_16U),
+      "AVIF only supports 1, 3, 4 channels and CV_8U and CV_16U");
+
+  Mat read_img;
+  if (img.channels() == channels_) {
+    read_img = img;
+  } else {
+    // Use the asked depth but keep the number of channels. OpenCV and not
+    // libavif will do the color conversion.
+    read_img.create(m_height, m_width, CV_MAKE_TYPE(img.depth(), channels_));
+  }
+
+  if (is_first_image_) {
+    if (!nextPage()) return false;
+    is_first_image_ = false;
+  }
+
+  if (CopyToMat(decoder_->image, channels_, &read_img) != AVIF_RESULT_OK) {
+    CV_Error(Error::StsInternal, "Cannot convert from AVIF to Mat");
+    return false;
+  }
+
+  if (decoder_->image->exif.size > 0) {
+    m_exif.parseExif(decoder_->image->exif.data, decoder_->image->exif.size);
+  }
+
+  if (img.channels() == channels_) {
+    // We already wrote to the right buffer.
+  } else {
+    if (channels_ == 1 && img.channels() == 3) {
+      cvtColor(read_img, img, COLOR_GRAY2BGR);
+    } else if (channels_ == 1 && img.channels() == 4) {
+      cvtColor(read_img, img, COLOR_GRAY2BGRA);
+    } else if (channels_ == 3 && img.channels() == 1) {
+      cvtColor(read_img, img, COLOR_BGR2GRAY);
+    } else if (channels_ == 3 && img.channels() == 4) {
+      cvtColor(read_img, img, COLOR_BGR2BGRA);
+    } else if (channels_ == 4 && img.channels() == 1) {
+      cvtColor(read_img, img, COLOR_BGRA2GRAY);
+    } else if (channels_ == 4 && img.channels() == 3) {
+      cvtColor(read_img, img, COLOR_BGRA2BGR);
+    } else {
+      CV_Error(Error::StsInternal, "");
+    }
+  }
+  return true;
+}
+
+bool AvifDecoder::nextPage() {
+  const avifResult status = avifDecoderNextImage(decoder_);
+  if (status == AVIF_RESULT_NO_IMAGES_REMAINING) return false;
+  if (status != AVIF_RESULT_OK) {
+    const std::string error(decoder_->diag.error);
+    CV_Error(Error::StsParseError, error + " " + avifResultToString(status));
+    return false;
+  }
+  return true;
+}
+
+////////////////////////////////////////////////////////////////////////////////
+
+AvifEncoder::AvifEncoder() {
+  m_description = "AVIF files (*.avif)";
+  m_buf_supported = true;
+  encoder_ = avifEncoderCreate();
+}
+
+AvifEncoder::~AvifEncoder() {
+  if (encoder_) avifEncoderDestroy(encoder_);
+}
+
+bool AvifEncoder::isFormatSupported(int depth) const {
+  return (depth == CV_8U || depth == CV_16U);
+}
+
+bool AvifEncoder::write(const Mat &img, const std::vector<int> &params) {
+  std::vector<Mat> img_vec(1, img);
+  return writeToOutput(img_vec, params);
+}
+
+bool AvifEncoder::writemulti(const std::vector<Mat> &img_vec,
+                             const std::vector<int> &params) {
+  return writeToOutput(img_vec, params);
+}
+
+bool AvifEncoder::writeToOutput(const std::vector<Mat> &img_vec,
+                                const std::vector<int> &params) {
+  int bit_depth = 8;
+  int speed = AVIF_SPEED_FASTEST;
+  for (size_t i = 0; i < params.size(); i += 2) {
+    if (params[i] == IMWRITE_AVIF_QUALITY) {
+      const int quality = std::min(std::max(params[i + 1], AVIF_QUALITY_WORST),
+                                   AVIF_QUALITY_BEST);
+#if CV_AVIF_USE_QUALITY
+      encoder_->quality = quality;
+#else
+      encoder_->minQuantizer = encoder_->maxQuantizer =
+          (AVIF_QUANTIZER_BEST_QUALITY - AVIF_QUANTIZER_WORST_QUALITY) *
+              quality / (AVIF_QUALITY_BEST - AVIF_QUALITY_WORST) +
+          AVIF_QUANTIZER_WORST_QUALITY;
+#endif
+    } else if (params[i] == IMWRITE_AVIF_DEPTH) {
+      bit_depth = params[i + 1];
+    } else if (params[i] == IMWRITE_AVIF_SPEED) {
+      speed = params[i + 1];
+    }
+  }
+
+  avifRWData output_ori = AVIF_DATA_EMPTY;
+  std::unique_ptr<avifRWData, decltype(&avifRWDataFree)> output(&output_ori,
+                                                                avifRWDataFree);
+#if CV_AVIF_USE_QUALITY
+  const bool do_lossless = (encoder_->quality == AVIF_QUALITY_LOSSLESS);
+#else
+  const bool do_lossless =
+      (encoder_->minQuantizer == AVIF_QUANTIZER_BEST_QUALITY &&
+       encoder_->maxQuantizer == AVIF_QUANTIZER_BEST_QUALITY);
+#endif
+  encoder_->speed = speed;
+
+  const avifAddImageFlags flag = (img_vec.size() == 1)
+                                     ? AVIF_ADD_IMAGE_FLAG_SINGLE
+                                     : AVIF_ADD_IMAGE_FLAG_NONE;
+  std::vector<AvifImageUniquePtr> images;
+  std::vector<cv::Mat> imgs_scaled;
+  for (const cv::Mat &img : img_vec) {
+    CV_CheckType(
+        img.type(),
+        (bit_depth == 8 && img.depth() == CV_8U) ||
+            ((bit_depth == 10 || bit_depth == 12) && img.depth() == CV_16U),
+        "AVIF only supports bit depth of 8 with CV_8U input or "
+        "bit depth of 10 or 12 with CV_16U input");
+    CV_Check(img.channels(),
+             img.channels() == 1 || img.channels() == 3 || img.channels() == 4,
+             "AVIF only supports 1, 3, 4 channels");
+
+    images.emplace_back(ConvertToAvif(img, do_lossless, bit_depth));
+  }
+  for (const AvifImageUniquePtr &image : images) {
+    OPENCV_AVIF_CHECK_STATUS(
+        avifEncoderAddImage(encoder_, image.get(), /*durationInTimescale=*/1,
+                            flag),
+        encoder_);
+  }
+
+  OPENCV_AVIF_CHECK_STATUS(avifEncoderFinish(encoder_, output.get()), encoder_);
+
+  if (m_buf) {
+    m_buf->resize(output->size);
+    std::memcpy(m_buf->data(), output->data, output->size);
+  } else {
+    std::ofstream(m_filename, std::ofstream::binary)
+        .write(reinterpret_cast<char *>(output->data), output->size);
+  }
+
+  return (output->size > 0);
+}
+
+ImageEncoder AvifEncoder::newEncoder() const { return makePtr<AvifEncoder>(); }
+
+}  // namespace cv
+
+#endif

modules/imgcodecs/src/grfmt_avif.hpp

@@ -0,0 +1,62 @@
+// This file is part of OpenCV project.
+// It is subject to the license terms in the LICENSE file found in the top-level
+// directory of this distribution and at http://opencv.org/license.html
+
+#ifndef _GRFMT_AVIF_H_
+#define _GRFMT_AVIF_H_
+
+#include "grfmt_base.hpp"
+
+#ifdef HAVE_AVIF
+
+struct avifDecoder;
+struct avifEncoder;
+struct avifRWData;
+
+namespace cv {
+
+class AvifDecoder CV_FINAL : public BaseImageDecoder {
+ public:
+  AvifDecoder();
+  ~AvifDecoder();
+
+  bool readHeader() CV_OVERRIDE;
+  bool readData(Mat& img) CV_OVERRIDE;
+  bool nextPage() CV_OVERRIDE;
+
+  size_t signatureLength() const CV_OVERRIDE;
+  bool checkSignature(const String& signature) const CV_OVERRIDE;
+  ImageDecoder newDecoder() const CV_OVERRIDE;
+
+ protected:
+  int channels_;
+  int bit_depth_;
+  avifDecoder* decoder_;
+  bool is_first_image_;
+};
+
+class AvifEncoder CV_FINAL : public BaseImageEncoder {
+ public:
+  AvifEncoder();
+  ~AvifEncoder() CV_OVERRIDE;
+
+  bool isFormatSupported(int depth) const CV_OVERRIDE;
+
+  bool write(const Mat& img, const std::vector<int>& params) CV_OVERRIDE;
+
+  bool writemulti(const std::vector<Mat>& img_vec,
+                  const std::vector<int>& params) CV_OVERRIDE;
+
+  ImageEncoder newEncoder() const CV_OVERRIDE;
+
+ private:
+  bool writeToOutput(const std::vector<Mat>& img_vec,
+                     const std::vector<int>& params);
+  avifEncoder* encoder_;
+};
+
+}  // namespace cv
+
+#endif
+
+#endif /*_GRFMT_AVIF_H_*/

modules/imgcodecs/src/grfmts.hpp

@@ -43,6 +43,7 @@
 #define _GRFMTS_H_
 
 #include "grfmt_base.hpp"
+#include "grfmt_avif.hpp"
 #include "grfmt_bmp.hpp"
 #include "grfmt_sunras.hpp"
 #include "grfmt_jpeg.hpp"

modules/imgcodecs/src/loadsave.cpp

@@ -132,6 +132,10 @@ struct ImageCodecInitializer
     */
     ImageCodecInitializer()
     {
+#ifdef HAVE_AVIF
+        decoders.push_back(makePtr<AvifDecoder>());
+        encoders.push_back(makePtr<AvifEncoder>());
+#endif
         /// BMP Support
         decoders.push_back( makePtr<BmpDecoder>() );
         encoders.push_back( makePtr<BmpEncoder>() );

modules/imgcodecs/test/test_avif.cpp

@@ -0,0 +1,355 @@
+// This file is part of OpenCV project.
+// It is subject to the license terms in the LICENSE file found in the top-level
+// directory of this distribution and at http://opencv.org/license.html
+
+#include <cstdint>
+#include <fstream>
+
+#include "test_precomp.hpp"
+
+#ifdef HAVE_AVIF
+
+namespace opencv_test {
+namespace {
+
+class Imgcodecs_Avif_RoundTripSuite
+    : public testing::TestWithParam<std::tuple<int, int, int, ImreadModes>> {
+ protected:
+  static cv::Mat modifyImage(const cv::Mat& img_original, int channels,
+                             int bit_depth) {
+    cv::Mat img;
+    if (channels == 1) {
+      cv::cvtColor(img_original, img, cv::COLOR_BGR2GRAY);
+    } else if (channels == 4) {
+      std::vector<cv::Mat> imgs;
+      cv::split(img_original, imgs);
+      imgs.push_back(cv::Mat(imgs[0]));
+      imgs[imgs.size() - 1] = cv::Scalar::all(128);
+      cv::merge(imgs, img);
+    } else {
+      img = img_original.clone();
+    }
+
+    cv::Mat img_final = img;
+    // Convert image to CV_16U for some bit depths.
+    if (bit_depth > 8) img.convertTo(img_final, CV_16U, 1 << (bit_depth - 8));
+
+    return img_final;
+  }
+
+  void SetUp() {
+    bit_depth_ = std::get<0>(GetParam());
+    channels_ = std::get<1>(GetParam());
+    quality_ = std::get<2>(GetParam());
+    imread_mode_ = std::get<3>(GetParam());
+    encoding_params_ = {cv::IMWRITE_AVIF_QUALITY, quality_,
+                        cv::IMWRITE_AVIF_DEPTH, bit_depth_};
+  }
+
+  bool IsBitDepthValid() const {
+    return (bit_depth_ == 8 || bit_depth_ == 10 || bit_depth_ == 12);
+  }
+
+  // Makes sure images are close enough after encode/decode roundtrip.
+  void ValidateRead(const cv::Mat& img_original, const cv::Mat& img) const {
+    EXPECT_EQ(img_original.size(), img.size());
+    if (imread_mode_ == IMREAD_UNCHANGED) {
+      ASSERT_EQ(img_original.type(), img.type());
+      // Lossless.
+      if (quality_ == 100) {
+        EXPECT_EQ(0, cvtest::norm(img, img_original, NORM_INF));
+      } else {
+        const float norm = cvtest::norm(img, img_original, NORM_L2) /
+                           img.channels() / img.cols / img.rows /
+                           (1 << (bit_depth_ - 8));
+        if (quality_ == 50) {
+          EXPECT_LE(norm, 10);
+        } else if (quality_ == 0) {
+          EXPECT_LE(norm, 13);
+        } else {
+          EXPECT_FALSE(true);
+        }
+      }
+    }
+  }
+
+ public:
+  int bit_depth_;
+  int channels_;
+  int quality_;
+  int imread_mode_;
+  std::vector<int> encoding_params_;
+};
+
+////////////////////////////////////////////////////////////////////////////////
+
+class Imgcodecs_Avif_Image_RoundTripSuite
+    : public Imgcodecs_Avif_RoundTripSuite {
+ public:
+  const cv::Mat& get_img_original() {
+    const Key key = {channels_, (bit_depth_ < 8) ? 8 : bit_depth_};
+    return imgs_[key];
+  }
+
+  // Prepare the original image modified for different number of channels and
+  // bit depth.
+  static void SetUpTestCase() {
+    const string root = cvtest::TS::ptr()->get_data_path();
+    const string filename = root + "../cv/shared/lena.png";
+    const cv::Mat img_original = cv::imread(filename);
+    cv::Mat img_resized;
+    cv::resize(img_original, img_resized, cv::Size(kWidth, kHeight), 0, 0);
+    for (int channels : {1, 3, 4}) {
+      for (int bit_depth : {8, 10, 12}) {
+        const Key key{channels, bit_depth};
+        imgs_[key] = modifyImage(img_resized, channels, bit_depth);
+      }
+    }
+  }
+
+  static const int kWidth;
+  static const int kHeight;
+
+ private:
+  typedef std::tuple<int, int> Key;
+  static std::map<Key, cv::Mat> imgs_;
+};
+std::map<std::tuple<int, int>, cv::Mat>
+    Imgcodecs_Avif_Image_RoundTripSuite::imgs_;
+const int Imgcodecs_Avif_Image_RoundTripSuite::kWidth = 51;
+const int Imgcodecs_Avif_Image_RoundTripSuite::kHeight = 31;
+
+class Imgcodecs_Avif_Image_WriteReadSuite
+    : public Imgcodecs_Avif_Image_RoundTripSuite {};
+
+TEST_P(Imgcodecs_Avif_Image_WriteReadSuite, imwrite_imread) {
+  const cv::Mat& img_original = get_img_original();
+  ASSERT_FALSE(img_original.empty());
+
+  // Encode.
+  const string output = cv::tempfile(".avif");
+  if (!IsBitDepthValid()) {
+    EXPECT_NO_FATAL_FAILURE(
+        cv::imwrite(output, img_original, encoding_params_));
+    EXPECT_NE(0, remove(output.c_str()));
+    return;
+  }
+  EXPECT_NO_THROW(cv::imwrite(output, img_original, encoding_params_));
+
+  // Read from file.
+  const cv::Mat img = cv::imread(output, imread_mode_);
+
+  ValidateRead(img_original, img);
+
+  EXPECT_EQ(0, remove(output.c_str()));
+}
+
+INSTANTIATE_TEST_CASE_P(
+    Imgcodecs_AVIF, Imgcodecs_Avif_Image_WriteReadSuite,
+    ::testing::Combine(::testing::ValuesIn({6, 8, 10, 12}),
+                       ::testing::ValuesIn({1, 3, 4}),
+                       ::testing::ValuesIn({0, 50, 100}),
+                       ::testing::ValuesIn({IMREAD_UNCHANGED, IMREAD_GRAYSCALE,
+                                            IMREAD_COLOR})));
+
+class Imgcodecs_Avif_Image_EncodeDecodeSuite
+    : public Imgcodecs_Avif_Image_RoundTripSuite {};
+
+TEST_P(Imgcodecs_Avif_Image_EncodeDecodeSuite, imencode_imdecode) {
+  const cv::Mat& img_original = get_img_original();
+  ASSERT_FALSE(img_original.empty());
+
+  // Encode.
+  std::vector<unsigned char> buf;
+  if (!IsBitDepthValid()) {
+    EXPECT_THROW(cv::imencode(".avif", img_original, buf, encoding_params_),
+                 cv::Exception);
+    return;
+  }
+  bool result;
+  EXPECT_NO_THROW(
+      result = cv::imencode(".avif", img_original, buf, encoding_params_););
+  EXPECT_TRUE(result);
+
+  // Read back.
+  const cv::Mat img = cv::imdecode(buf, imread_mode_);
+
+  ValidateRead(img_original, img);
+}
+
+INSTANTIATE_TEST_CASE_P(
+    Imgcodecs_AVIF, Imgcodecs_Avif_Image_EncodeDecodeSuite,
+    ::testing::Combine(::testing::ValuesIn({6, 8, 10, 12}),
+                       ::testing::ValuesIn({1, 3, 4}),
+                       ::testing::ValuesIn({0, 50, 100}),
+                       ::testing::ValuesIn({IMREAD_UNCHANGED, IMREAD_GRAYSCALE,
+                                            IMREAD_COLOR})));
+
+////////////////////////////////////////////////////////////////////////////////
+
+typedef testing::TestWithParam<string> Imgcodecs_AVIF_Exif;
+
+TEST_P(Imgcodecs_AVIF_Exif, exif_orientation) {
+  const string root = cvtest::TS::ptr()->get_data_path();
+  const string filename = root + GetParam();
+  const int colorThresholdHigh = 250;
+  const int colorThresholdLow = 5;
+
+  Mat m_img = imread(filename);
+  ASSERT_FALSE(m_img.empty());
+  Vec3b vec;
+
+  // Checking the first quadrant (with supposed red)
+  vec = m_img.at<Vec3b>(2, 2);  // some point inside the square
+  EXPECT_LE(vec.val[0], colorThresholdLow);
+  EXPECT_LE(vec.val[1], colorThresholdLow);
+  EXPECT_GE(vec.val[2], colorThresholdHigh);
+
+  // Checking the second quadrant (with supposed green)
+  vec = m_img.at<Vec3b>(2, 7);  // some point inside the square
+  EXPECT_LE(vec.val[0], colorThresholdLow);
+  EXPECT_GE(vec.val[1], colorThresholdHigh);
+  EXPECT_LE(vec.val[2], colorThresholdLow);
+
+  // Checking the third quadrant (with supposed blue)
+  vec = m_img.at<Vec3b>(7, 2);  // some point inside the square
+  EXPECT_GE(vec.val[0], colorThresholdHigh);
+  EXPECT_LE(vec.val[1], colorThresholdLow);
+  EXPECT_LE(vec.val[2], colorThresholdLow);
+}
+
+const string exif_files[] = {"readwrite/testExifOrientation_1.avif",
+                             "readwrite/testExifOrientation_2.avif",
+                             "readwrite/testExifOrientation_3.avif",
+                             "readwrite/testExifOrientation_4.avif",
+                             "readwrite/testExifOrientation_5.avif",
+                             "readwrite/testExifOrientation_6.avif",
+                             "readwrite/testExifOrientation_7.avif",
+                             "readwrite/testExifOrientation_8.avif"};
+
+INSTANTIATE_TEST_CASE_P(ExifFiles, Imgcodecs_AVIF_Exif,
+                        testing::ValuesIn(exif_files));
+
+////////////////////////////////////////////////////////////////////////////////
+
+class Imgcodecs_Avif_Animation_RoundTripSuite
+    : public Imgcodecs_Avif_RoundTripSuite {
+ public:
+  const std::vector<cv::Mat>& get_anim_original() {
+    const Key key = {channels_, bit_depth_};
+    return anims_[key];
+  }
+
+  // Prepare the original image modified for different number of channels and
+  // bit depth.
+  static void SetUpTestCase() {
+    const string root = cvtest::TS::ptr()->get_data_path();
+    const string filename = root + "../cv/shared/lena.png";
+    const cv::Mat img_original = cv::imread(filename);
+    cv::Mat img_resized;
+    cv::resize(img_original, img_resized, cv::Size(kWidth, kHeight), 0, 0);
+    for (int channels : {1, 3, 4}) {
+      for (int bit_depth : {8, 10, 12}) {
+        const Key key{channels, bit_depth};
+        const cv::Mat img = modifyImage(img_resized, channels, bit_depth);
+        cv::Mat img2, img3;
+        cv::flip(img, img2, 0);
+        cv::flip(img, img3, -1);
+        anims_[key] = {img, img2, img3};
+      }
+    }
+  }
+
+  void ValidateRead(const std::vector<cv::Mat>& anim_original,
+                    const std::vector<cv::Mat>& anim) const {
+    ASSERT_EQ(anim_original.size(), anim.size());
+    for (size_t i = 0; i < anim.size(); ++i) {
+      Imgcodecs_Avif_RoundTripSuite::ValidateRead(anim_original[i], anim[i]);
+    }
+  }
+
+  static const int kWidth;
+  static const int kHeight;
+
+ private:
+  typedef std::tuple<int, int> Key;
+  static std::map<Key, std::vector<cv::Mat>> anims_;
+};
+std::map<std::tuple<int, int>, std::vector<cv::Mat>>
+    Imgcodecs_Avif_Animation_RoundTripSuite::anims_;
+const int Imgcodecs_Avif_Animation_RoundTripSuite::kWidth = 5;
+const int Imgcodecs_Avif_Animation_RoundTripSuite::kHeight = 5;
+
+class Imgcodecs_Avif_Animation_WriteReadSuite
+    : public Imgcodecs_Avif_Animation_RoundTripSuite {};
+
+TEST_P(Imgcodecs_Avif_Animation_WriteReadSuite, encode_decode) {
+  const std::vector<cv::Mat>& anim_original = get_anim_original();
+  ASSERT_FALSE(anim_original.empty());
+
+  // Encode.
+  const string output = cv::tempfile(".avif");
+  if (!IsBitDepthValid()) {
+    EXPECT_THROW(cv::imwritemulti(output, anim_original, encoding_params_),
+                 cv::Exception);
+    EXPECT_NE(0, remove(output.c_str()));
+    return;
+  }
+  EXPECT_NO_THROW(cv::imwritemulti(output, anim_original, encoding_params_));
+
+  // Read from file.
+  std::vector<cv::Mat> anim;
+  ASSERT_TRUE(cv::imreadmulti(output, anim, imread_mode_));
+
+  ValidateRead(anim_original, anim);
+
+  EXPECT_EQ(0, remove(output.c_str()));
+}
+
+INSTANTIATE_TEST_CASE_P(
+    Imgcodecs_AVIF, Imgcodecs_Avif_Animation_WriteReadSuite,
+    ::testing::Combine(::testing::ValuesIn({8, 10, 12}),
+                       ::testing::ValuesIn({1, 3}), ::testing::ValuesIn({50}),
+                       ::testing::ValuesIn({IMREAD_UNCHANGED, IMREAD_GRAYSCALE,
+                                            IMREAD_COLOR})));
+class Imgcodecs_Avif_Animation_WriteDecodeSuite
+    : public Imgcodecs_Avif_Animation_RoundTripSuite {};
+
+TEST_P(Imgcodecs_Avif_Animation_WriteDecodeSuite, encode_decode) {
+  const std::vector<cv::Mat>& anim_original = get_anim_original();
+  ASSERT_FALSE(anim_original.empty());
+
+  // Encode.
+  const string output = cv::tempfile(".avif");
+  if (!IsBitDepthValid()) {
+    EXPECT_THROW(cv::imwritemulti(output, anim_original, encoding_params_),
+                 cv::Exception);
+    EXPECT_NE(0, remove(output.c_str()));
+    return;
+  }
+  EXPECT_NO_THROW(cv::imwritemulti(output, anim_original, encoding_params_));
+
+  // Put file into buffer and read from buffer.
+  std::ifstream file(output, std::ios::binary | std::ios::ate);
+  std::streamsize size = file.tellg();
+  file.seekg(0, std::ios::beg);
+  std::vector<unsigned char> buf(size);
+  EXPECT_TRUE(file.read(reinterpret_cast<char*>(buf.data()), size));
+  EXPECT_EQ(0, remove(output.c_str()));
+  std::vector<cv::Mat> anim;
+  ASSERT_TRUE(cv::imdecodemulti(buf, imread_mode_, anim));
+
+  ValidateRead(anim_original, anim);
+}
+
+INSTANTIATE_TEST_CASE_P(
+    Imgcodecs_AVIF, Imgcodecs_Avif_Animation_WriteDecodeSuite,
+    ::testing::Combine(::testing::ValuesIn({8, 10, 12}),
+                       ::testing::ValuesIn({1, 3}), ::testing::ValuesIn({50}),
+                       ::testing::ValuesIn({IMREAD_UNCHANGED, IMREAD_GRAYSCALE,
+                                            IMREAD_COLOR})));
+
+}  // namespace
+}  // namespace opencv_test
+
+#endif  // HAVE_AVIF