Simd downsampling

.github/workflows/main.yml

@@ -14,6 +14,8 @@ jobs:
     steps:
       - name: Checkout repository
         uses: actions/checkout@v4
+        with:
+          submodules: true
 
       - name: Install dependencies (Linux)
         if: runner.os == 'Linux'
@@ -31,4 +33,4 @@ jobs:
         run: brew install gcc make
 
       - name: Build
-        run: git clone https://github.com/libjpeg-turbo/libjpeg-turbo.git && ./build.sh  build macos
+        run: ./build.sh  build macos

.gitignore

@@ -1,5 +1,5 @@
 !files/apple.jpeg
-!files/orange.jpeg 
+!files/orange.jpeg
 files/*
 *.jpeg
 *.jpg
@@ -15,4 +15,6 @@ installs/*
 tests/test
 *.dSYM*
 downsampled
-test
+test
+example.c
+example

.gitmodules

@@ -1,3 +1,6 @@
 [submodule "libjpeg-turbo"]
 	path = libjpeg-turbo
 	url = https://github.com/libjpeg-turbo/libjpeg-turbo.git
+[submodule "simde"]
+	path = simde
+	url = https://github.com/simd-everywhere/simde.git

CMakeLists.txt

@@ -2,24 +2,45 @@ cmake_minimum_required(VERSION 2.8.12...3.28)
 
 project(NativeSticher LANGUAGES C)
 
-set(LIBJPEG_TURBO_ROOT "${LIBJPEG_TURBO_ROOT}") 
+set(LIBJPEG_TURBO_ROOT      "${LIBJPEG_TURBO_ROOT}")
 set(LIBJPEG_TURBO_INCLUDE_DIR "${LIBJPEG_TURBO_ROOT}/include")
-set(LIBJPEG_TURBO_LIB_DIR "${LIBJPEG_TURBO_ROOT}/lib")
+set(LIBJPEG_TURBO_LIB_DIR     "${LIBJPEG_TURBO_ROOT}/lib")
 set(LIBJPEG_LIBS turbojpeg)
 
 if(CMAKE_INSTALL_PREFIX_INITIALIZED_TO_DEFAULT)
-    message(FATAL_ERROR "CMAKE_INSTALL_PREFIX not set. Please provide a prefix using -DCMAKE_INSTALL_PREFIX=<path>")
+    message(FATAL_ERROR
+      "CMAKE_INSTALL_PREFIX not set. "
+      "Please provide a prefix using -DCMAKE_INSTALL_PREFIX=<path>")
 endif()
 
 
-include_directories(${LIBJPEG_TURBO_INCLUDE_DIR})
+include_directories(
+    ${LIBJPEG_TURBO_INCLUDE_DIR}
+     ${CMAKE_SOURCE_DIR}/simde
+)
 link_directories(${LIBJPEG_TURBO_LIB_DIR})
 
-add_library(${PROJECT_NAME} SHARED laplace_blending.c jpeg.c utils.c)
+add_library(${PROJECT_NAME} SHARED
+    image_operations.c
+    laplace_blending.c
+    jpeg.c
+    utils.c
+)
+
+target_compile_options(${PROJECT_NAME} PRIVATE -O3 -pthread)
+target_link_libraries(${PROJECT_NAME} PRIVATE -pthread)
+target_compile_definitions(${PROJECT_NAME} PRIVATE SIMDE_ENABLE_NATIVE_ALIASES)
+
 target_link_libraries(${PROJECT_NAME} PRIVATE ${LIBJPEG_LIBS})
 
+install(TARGETS ${PROJECT_NAME}
+        LIBRARY DESTINATION lib)
 
-install(TARGETS ${PROJECT_NAME} LIBRARY DESTINATION lib)
-install(DIRECTORY ${LIBJPEG_TURBO_INCLUDE_DIR}/ DESTINATION include) 
-install(FILES laplace_blending.h utils.h jpeg.h  DESTINATION include)
+install(DIRECTORY ${LIBJPEG_TURBO_INCLUDE_DIR}/
+        DESTINATION include)
 
+install(FILES image_operations.h
+              laplace_blending.h
+              utils.h
+              jpeg.h
+        DESTINATION include)

README.md

@@ -1,40 +1,40 @@
 # NativeSticher
-**NativeSticher** is a C library for multiband image merging using the Laplacian blending technique.  
+**NativeSticher** is a C library for multiband image merging using the Laplacian blending technique.
 It enables seamless blending of overlapping images, useful for applications such as panorama generation, image stitching, and exposure fusion. Only supports JPEGS for now.
 
 ---
 
 ## 🚧 Building
-A helper script (`build.sh`) is included to streamline the build process for different platforms.  
-Use this script to build or clean the project easily.  
+A helper script (`build.sh`) is included to streamline the build process for different platforms.
+Use this script to build or clean the project easily.
 
 ### 🔨 Build Instructions:
-- **Build for macOS:**  
+- **Build for macOS:**
   ```bash
   ./build.sh build macos
   ```
-- **Build for ios:**  
+- **Build for ios:**
   ```bash
   ./build.sh build ios
   ```
 
-# 🧪 Testing  
+# 🧪 Testing
 
-To verify the functionality of **NativeSticher**, follow the instructions below based on your setup.  
+To verify the functionality of **NativeSticher**, follow the instructions below based on your setup.
 
 ---
 
-### 1. Testing with libturbojpeg (Direct Compilation)  
-If you have `libturbojpeg` installed, compile and run the test with the following command:  
+### 1. Testing with libturbojpeg (Direct Compilation)
+If you have `libturbojpeg` installed, compile and run the test with the following command:
 ```bash
-gcc-14 -pthread -fsanitize=address -g -o stitch \
+gcc-14 -O3 -mavx2 -mfma -I simde/ -pthread -fsanitize=address -g -o stitch \
 -I../ -I/usr/local/include \
 -L/usr/local/lib -lturbojpeg \
-stitch.c ../laplace_blending.c ../jpeg.c ../utils.c && time ./stitch
+stitch.c ../laplace_blending.c ../jpeg.c ../image_operations.c ../utils.c && time ./stitch
 ```
 
-### 2. Testing with Custom-Built NativeSticher Library  
-If you have manually built the NativeSticher library, use the following command to test it:  
+### 2. Testing with Custom-Built NativeSticher Library
+If you have manually built the NativeSticher library, use the following command to test it:
 ```bash
 gcc-14 -pthread -fsanitize=address -g -o stitch \
 -I../installs/native-stitcher/macos/x86_64/include \
@@ -43,5 +43,3 @@ gcc-14 -pthread -fsanitize=address -g -o stitch \
 -lNativeSticher stitch.c && ./stitch
 ```
 The commands above assume a mac as working machine.
-
-

build.sh

@@ -1,5 +1,5 @@
 #!/bin/bash
-set -e 
+set -e
 BUILD_DIR_LIB_TURBOJPEG=$PWD/builds/libturbojpeg
 INSTALL_DIR_LIB_TURBOJPEG=$PWD/installs/libturbojpeg
 BUILD_DIR_LIB_NATIVE_STITCHER=$PWD/builds/native-stitcher
@@ -34,6 +34,7 @@ build_macos() {
       -DCMAKE_SYSTEM_PROCESSOR=$ARCH \
       -DCMAKE_INSTALL_PREFIX="$INSTALL_DIR" \
       -DCMAKE_C_COMPILER=$(xcrun --sdk macosx --find clang) \
+      -DCMAKE_OSX_SYSROOT=$(xcrun --sdk macosx --show-sdk-path) \
       -DCMAKE_C_FLAGS="$TARGET_FLAG"
 
     make -j$(sysctl -n hw.logicalcpu)
@@ -52,10 +53,12 @@ build_macos() {
         -DCMAKE_SYSTEM_PROCESSOR=$ARCH \
         -DCMAKE_INSTALL_PREFIX="$INSTALL_DIR" \
         -DCMAKE_C_COMPILER=$(xcrun --sdk macosx --find clang) \
+        -DCMAKE_OSX_ARCHITECTURES=$ARCH \
+        -DCMAKE_OSX_SYSROOT=$(xcrun --sdk macosx --show-sdk-path) \
         -DCMAKE_C_FLAGS="$TARGET_FLAG"
       make -j$(sysctl -n hw.logicalcpu)
       make install
-    popd    
+    popd
   done
 }
 
@@ -65,7 +68,7 @@ build_android() {
     BUILD_DIR="$BUILD_DIR_LIB_TURBOJPEG/andriod/$ARCH"
     INSTALL_DIR="$INSTALL_DIR_LIB_TURBOJPEG/andriod/$ARCH"
     mkdir -p "$BUILD_DIR"
-    
+
 
     if [ "$ARCH" == "arm64-v8a" ]; then
       TARGET_FLAG="-target aarch64-linux-android21"
@@ -120,7 +123,7 @@ build_android() {
         -DCMAKE_INSTALL_PREFIX="$INSTALL_DIR" \
         -DCMAKE_C_COMPILER=$ANDROID_NDK/toolchains/llvm/prebuilt/darwin-x86_64/bin/clang \
         -DCMAKE_C_FLAGS="$TARGET_FLAG" \
-        -DCMAKE_TOOLCHAIN_FILE=$ANDROID_NDK/build/cmake/android.toolchain.cmake 
+        -DCMAKE_TOOLCHAIN_FILE=$ANDROID_NDK/build/cmake/android.toolchain.cmake
       make -j$(sysctl -n hw.logicalcpu)
       make install
     popd
@@ -148,7 +151,7 @@ build_ios() {
         -DCMAKE_OSX_SYSROOT=${IOS_SYSROOT[0]} \
         -DCMAKE_INSTALL_PREFIX="$INSTALL_DIR" \
         -DCMAKE_C_COMPILER=$(xcrun --sdk iphoneos --find clang) \
-        -DCMAKE_C_FLAGS="$TARGET_FLAG" 
+        -DCMAKE_C_FLAGS="$TARGET_FLAG"
 
       make -j$(sysctl -n hw.logicalcpu)
       make install
@@ -167,7 +170,7 @@ build_ios() {
         -DCMAKE_OSX_SYSROOT=${IOS_SYSROOT[0]} \
         -DCMAKE_C_FLAGS="-target arm64-apple-ios12.0" \
         -DCMAKE_INSTALL_PREFIX="$INSTALL_DIR" \
-        -DCMAKE_C_COMPILER=$(xcrun --sdk iphoneos --find clang) 
+        -DCMAKE_C_COMPILER=$(xcrun --sdk iphoneos --find clang)
       make -j$(sysctl -n hw.logicalcpu)
       make install
     popd
@@ -193,7 +196,7 @@ build_ios_sim() {
           -DCMAKE_OSX_SYSROOT=${IOS_SYSROOT[0]} \
           -DCMAKE_INSTALL_PREFIX="$INSTALL_DIR" \
           -DCMAKE_C_COMPILER=$(xcrun --sdk iphoneos --find clang) \
-          -DCMAKE_C_FLAGS="$TARGET_FLAG" 
+          -DCMAKE_C_FLAGS="$TARGET_FLAG"
 
         make -j$(sysctl -n hw.logicalcpu)
         make install
@@ -212,7 +215,7 @@ build_ios_sim() {
           -DCMAKE_OSX_SYSROOT=${IOS_SYSROOT[0]} \
           -DCMAKE_INSTALL_PREFIX="$INSTALL_DIR" \
           -DCMAKE_C_FLAGS="-target $ARCH-apple-ios14.0-simulator" \
-          -DCMAKE_C_COMPILER=$(xcrun --sdk iphonesimulator --find clang) 
+          -DCMAKE_C_COMPILER=$(xcrun --sdk iphonesimulator --find clang)
         make -j$(sysctl -n hw.logicalcpu)
         make install
       popd
@@ -269,7 +272,7 @@ case "$1" in
             ;;
         ios-sim)
             build_ios_sim
-            ;;              
+            ;;
         *)
         help
         ;;
@@ -285,4 +288,4 @@ esac
 
 
 # gcc-14 -I../ -I/usr/local/include -L/usr/local/lib -lturbojpeg -pthread -fsanitize=address -g -o downsampled ../laplace_blending.c ../jpeg.c ../utils.c downsampling.c && ./downsampled
-# gcc-14 -I../ -I/usr/local/include -L/usr/local/lib -lturbojpeg -pthread -fsanitize=address -g -o stitch  ../laplace_blending.c ../jpeg.c ../utils.c stitch.c && ./stitch
+# gcc-14 -I../ -I/usr/local/include -L/usr/local/lib -lturbojpeg -pthread -fsanitize=address -g -o stitch  ../laplace_blending.c ../jpeg.c ../utils.c stitch.c && ./stitch

examples/downsampling.c

@@ -25,7 +25,7 @@ int main()
 
     Image *img = &img_buf1;
     char buf[100];
-    
+
     for (int i = 0; i < 3; i++)
     {
         Image down = downsample(img);
@@ -53,11 +53,11 @@ int main()
         img->width = up.width;
         img->height = up.height;
     }
-    
-    
+
+
 
     destroy_image(&img_buf1);
     destroy_image(&down);
     destroy_image(&up);
     destroy_image(&mask1);
-}
+}