Skip to main content

6-2 Cross Compilation of OpenCV

This document outlines the steps to cross-compile OpenCV applications for ARM-based Renesas platforms using a Yocto SDK, including both C++ and Python implementations.

Environment Setup for Renesas Cross Compilation​

To begin, follow the instructions in Section 3-3 to build and install the SDK. By default, the SDK is installed at:

/opt/poky/3.1.31

To enable the cross-compilation environment, execute the following commands:

cd /opt/poky/3.1.31
unset LD_LIBRARY_PATH
. environment-setup-aarch64-poky-linux

To verify the setup, run:

echo $CXX

You should see an output like this:

aarch64-poky-linux-g++ -mtune=cortex-a55 -fstack-protector-strong -D_FORTIFY_SOURCE=2 ...

Cross-Compiling OpenCV for ARM-Based Linux Systems​

Refer to the following official OpenCV documentation:

OpenCV C++ Example Program and CMake Build​

Example Program: DisplayImage.cpp​

#include <stdio.h>
#include <opencv2/opencv.hpp>

using namespace cv;

int main(int argc, char* argv[]) {
  if (argc != 2) {
    printf("usage: DisplayImage.out <Image_Path>\n");
    return -1;
  }

  Mat image = imread(argv[1], 1);

  if (!image.data) {
    printf("No image data\n");
    return -1;
  }

  namedWindow("Display Image", WINDOW_AUTOSIZE);
  imshow("Display Image", image);
  waitKey(0);

  return 0;
}

CMake Configuration: CMakeLists.txt​

cmake_minimum_required(VERSION 2.8)
project(DisplayImage)

find_package(OpenCV REQUIRED)
add_executable(DisplayImage DisplayImage.cpp)
target_link_libraries(DisplayImage ${OpenCV_LIBS})

Sample Image​

Download Lenna.jpg from:

Lenna - Wikipedia

Building and Verifying Locally (x86_64)​

mkdir build
cd build
cmake ..
make
file ./DisplayImage

Expected output:

DisplayImage: ELF 64-bit LSB pie executable, x86-64 ...

Run the program:

./DisplayImage ../lena.jpg

Cross-Compiling for aarch64​

Step 1: Set Cross-Compilation Environment

cd /opt/poky/3.1.31
unset LD_LIBRARY_PATH
. environment-setup-aarch64-poky-linux

Step 2: Clean and Rebuild

cd /path/to/OpenCV/HelloWorld
rm -r build
mkdir build
cd build
cmake ..
make

Step 3: Verify Architecture

file ./DisplayImage

Expected output:

DisplayImage: ELF 64-bit LSB pie executable, ARM aarch64 ...

Running on the Renesas Board​

Transfer the DisplayImage binary to the Renesas board using the scp command and run it with:

./DisplayImage lena.jpg

Building with a Makefile​

Sample Makefile (System Installed OpenCV)

CPP = g++
CPPFLAGS = -L/usr/lib/x86_64-linux-gnu \
           -lopencv_core -lopencv_highgui \
           -lopencv_imgproc -lopencv_imgcodecs -lOpenCL \
           -I/usr/include/opencv4

all: DisplayImage

DisplayImage: DisplayImage.cpp
	$(CPP) $^ -o $@ $(CPPFLAGS)

Sample Makefile (Source-Installed OpenCV)

CPP = g++
CPPFLAGS = -L/usr/local/lib \
           -lopencv_core -lopencv_highgui \
           -lopencv_imgproc -lopencv_imgcodecs -lOpenCL \
           -I/usr/local/include/opencv4

all: DisplayImage

DisplayImage: DisplayImage.cpp
	$(CPP) $^ -o $@ $(CPPFLAGS)

To build:

make

Direct Compilation with g++​

Without pkg-config:

g++ -o DisplayImage DisplayImage.cpp \
-I/usr/local/include/opencv4 \
-L/usr/local/lib \
-lopencv_core -lopencv_highgui -lopencv_imgproc -lopencv_imgcodecs

With pkg-config: Check if opencv4.pc exists:

pkg-config --cflags opencv4

Compile:

g++ -o DisplayImage DisplayImage.cpp `pkg-config --cflags opencv4` `pkg-config --libs opencv4`

For Cross-Compilation with Yocto SDK:

$CXX -o DisplayImage DisplayImage.cpp `pkg-config --cflags opencv4` `pkg-config --libs opencv4`

OpenCV Python Example​

Python Script: DisplayImage.py

import cv2
import sys

if len(sys.argv) != 2:
    print("usage:", sys.argv[0], "<Image_Path>")
    sys.exit(-1)

img = cv2.imread(sys.argv[1], cv2.IMREAD_COLOR)
cv2.namedWindow("Display Image", cv2.WINDOW_AUTOSIZE)
cv2.imshow("Display Image", img)
cv2.waitKey(0)

Run:

python3 DisplayImage.py lena.jpg

Using Matplotlib to Display Image​

import cv2
import sys
from matplotlib import pyplot as plt

if len(sys.argv) != 2:
    print("usage:", sys.argv[0], "<Image_Path>")
    sys.exit(-1)

img = cv2.imread(sys.argv[1], cv2.IMREAD_COLOR)
plt.axis("off")
plt.imshow(cv2.cvtColor(img, cv2.COLOR_BGR2RGB))
plt.show()

Install Matplotlib:

pip install matplotlib