Opencv C++ Tutorial Mat resize

A comprehensive guide to scaling images, maintaining aspect ratios, and understanding interpolation.

Resizing an image is more than just changing its dimensions; it is a fundamental building block of Computer Vision. Whether you are preparing datasets for Deep Learning, optimizing web performance, or building a UI, mastering the cv::resize function in OpenCV is essential.

In this tutorial, we will move beyond the basics to explore interpolation algorithms and how to scale images without losing quality.

What You'll Learn:

• ✅ The cv::resize() Function Syntax

• ✅ Scaling by Fixed Dimensions vs. Scale Factors

• ✅ Maintaining Perfect Aspect Ratio

• ✅ Choosing the Right Interpolation (Area, Linear, Cubic)

• ✅ Full Implementation Code

The cv::resize Function

In OpenCV, resizing is handled by a single, powerful function. Here is the modern C++ signature:

void resize(InputArray src, OutputArray dst, Size dsize, 
            double fx = 0, double fy = 0, int interpolation = INTER_LINEAR);

src / dst: The source (input) and destination (output) image matrices.

dsize: The target size (width, height). If this is zero, fx and fy are used.

fx / fy: Scale factors (e.g., 0.5 for half size).

🛠️ Interactive Aspect Ratio Tool

Input your original image size to find the new dimensions while keeping the ratio.

x ➔ Target Width: New Height: -

Pro Tip: Choosing the Right Interpolation

Not all "resizes" are created equal. Depending on whether you are shrinking or enlarging, you should change the interpolation parameter:

Method Best For... INTER_NEAREST Fastest, but pixelated. Good for masks. INTER_LINEAR Default. Good balance of speed/quality. INTER_AREA Best for Downsampling (shrinking). Avoids moiré patterns. INTER_CUBIC Best for Upsampling (enlarging). Higher quality than linear.

Full C++ Implementation

This modernized code demonstrates how to load an image, display it, and perform different types of resizing.

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

using namespace cv;
using namespace std;

int main() {
    // 1. Load the original image
    Mat img = imread("Lenna.png", IMREAD_COLOR);
    if(img.empty()) {
        cout << "Could not read the image" << endl;
        return 1;
    }

    // 2. Simple Resize (Fixed Size: 100x100)
    Mat imgFixed;
    resize(img, imgFixed, Size(100, 100), 0, 0, INTER_LINEAR);

    // 3. Scale by Ratio (e.g., 50% of original)
    Mat imgScaled;
    resize(img, imgScaled, Size(), 0.5, 0.5, INTER_AREA);

    // 4. Maintain Aspect Ratio (Fixed Width of 300px)
    int newWidth = 300;
    double scale = (double)newWidth / img.cols;
    int newHeight = cvRound(img.rows * scale);
    Mat imgAspect;
    resize(img, imgAspect, Size(newWidth, newHeight), 0, 0, INTER_CUBIC);

    // Visualization
    imshow("Original", img);
    imshow("Fixed 100x100", imgFixed);
    imshow("Scaled 0.5x", imgScaled);
    imshow("Maintained Aspect", imgAspect);

    waitKey(0);
    return 0;
}

Visual Comparison: Lenna Resize

Original Mat

100x100 Fixed

0.5x Scale Ratio

Ready to level up your Computer Vision skills?

Resizing is just the beginning. Check out our other tutorials on Image Thresholding and Edge Detection to start building your own AI applications.

Explore More Tutorials