3.5.1 OpenCV Usage on RVV

Last Version: 10/09/2025

Using OpenCV on RVV (RISC-V Vector) is similar to using it on x86 platforms. The main difference is in how the libraries are built and optimized for RVV.

What is OpenCV?

OpenCV (Open Source Computer Vision Library) is an open-source computer vision and machine learning software library, initiated by Intel and widely supported by the community.
It provides a cross-platform programming framework for developing real-time computer vision applications.

Common use cases include:

• Image processing and analysis
• Face detection and recognition
• Object detection and tracking
• Machine learning applications
• Video analysis
• Camera calibration and 3D reconstruction

Basic OpenCV API Examples

Reading and Displaying Images

This example shows how to read, display, and save an image.

#include <opencv2/opencv.hpp>
#include <iostream>

int main() {
    // Read image
    cv::Mat image = cv::imread("input.jpg");

    if (image.empty()) {
        std::cout << "Failed to read image file" << std::endl;
        return -1;
    }

    // Display image
    cv::imshow("Original Image", image);
    cv::waitKey(0);

    // Save image
    cv::imwrite("output.jpg", image);

    return 0;
}

Basic Image Operations

This example demonstrates resizing and rotating images.

#include <opencv2/opencv.hpp>

int main() {
    cv::Mat image = cv::imread("input.jpg");

    // Get image information
    int width = image.cols;
    int height = image.rows;
    int channels = image.channels();

    std::cout << "Image dimensions: " << width << "x" << height << std::endl;
    std::cout << "Number of channels: " << channels << std::endl;

    // Resize image
    cv::Mat resized;
    cv::resize(image, resized, cv::Size(640, 480));

    // Rotate image
    cv::Point2f center(image.cols/2.0, image.rows/2.0);
    cv::Mat rotationMatrix = cv::getRotationMatrix2D(center, 45, 1.0);
    cv::Mat rotated;
    cv::warpAffine(image, rotated, rotationMatrix, image.size());

    cv::imshow("Resized", resized);
    cv::imshow("Rotated", rotated);
    cv::waitKey(0);

    return 0;
}

Color Space Conversion

This example converts an image to different color spaces.

#include <opencv2/opencv.hpp>

int main() {
    cv::Mat image = cv::imread("input.jpg");

    // Convert BGR to Grayscale
    cv::Mat gray;
    cv::cvtColor(image, gray, cv::COLOR_BGR2GRAY);

    // Convert BGR to HSV
    cv::Mat hsv;
    cv::cvtColor(image, hsv, cv::COLOR_BGR2HSV);

    // Convert BGR to LAB
    cv::Mat lab;
    cv::cvtColor(image, lab, cv::COLOR_BGR2LAB);

    cv::imshow("Original", image);
    cv::imshow("Grayscale", gray);
    cv::imshow("HSV", hsv);
    cv::imshow("LAB", lab);
    cv::waitKey(0);

    return 0;
}

Edge Detection

This example shows how to detect edges using Canny and Sobel methods.

#include <opencv2/opencv.hpp>

int main() {
    cv::Mat image = cv::imread("input.jpg", cv::IMREAD_GRAYSCALE);

    // Canny edge detection
    cv::Mat edges;
    cv::Canny(image, edges, 50, 150);

    // Sobel edge detection
    cv::Mat sobelX, sobelY, sobel;
    cv::Sobel(image, sobelX, CV_64F, 1, 0, 3);
    cv::Sobel(image, sobelY, CV_64F, 0, 1, 3);
    cv::magnitude(sobelX, sobelY, sobel);
    sobel.convertTo(sobel, CV_8U);

    cv::imshow("Original", image);
    cv::imshow("Canny Edges", edges);
    cv::imshow("Sobel Edges", sobel);
    cv::waitKey(0);

    return 0;
}

Performance Comparison

Comparing performance between RVV-accelerated OpenCV and non-RVV-accelerated OpenCV.

Performance test example:

#include <opencv2/opencv.hpp>
#include <iostream>
#include <chrono>

int main() {
    cv::Mat image = cv::imread("../data/test.jpg");

    auto start_time = std::chrono::steady_clock::now();
    cv::Mat blob;
    cv::resize(image, blob, cv::Size(640, 640));

    auto end_time = std::chrono::steady_clock::now();
    auto duration = std::chrono::duration_cast<std::chrono::milliseconds>(end_time - start_time);
    std::cout << "Resize time: " << duration.count() << " ms" << std::endl;

    start_time = std::chrono::steady_clock::now();
    cv::Mat blob_float;
    blob_float = cv::dnn::blobFromImage(image, 1.0 / 255.0, cv::Size(640, 640), cv::Scalar(0, 0, 0), false, false, CV_32F);
    end_time = std::chrono::steady_clock::now();
    duration = std::chrono::duration_cast<std::chrono::milliseconds>(end_time - start_time);
    std::cout << "blobFromImage time: " << duration.count() << " ms" << std::endl;

    return 0;
}

Performance Results

Operation	Non-RVV	RVV Optimized
Resize	10 ms	6 ms
blobFromImage	51 ms	35 ms

RVV optimization can significantly improve image processing performance.