askvity

How to Reduce the Resolution of an Image in Matlab?

Published in MATLAB Image Processing 4 mins read

You can reduce the resolution of an image in Matlab primarily using the imresize function.

Reducing image resolution, also known as downsampling or shrinking, effectively decreases the number of pixels in the image, resulting in a smaller file size and potentially faster processing, though quality may be reduced. Matlab's imresize function is the standard tool for this task.

Using the imresize Function

The imresize function allows you to change the spatial resolution of an image. You can specify the desired output size in several ways:

  1. Scaling Factor: Specify a scalar between 0 and 1 to shrink the image by that factor.
  2. Target Size: Specify the exact number of rows and columns for the output image.

The function also requires an interpolation method, which determines how the values of the new pixels are calculated from the original pixels.

Shrinking Using a Scaling Factor and Nearest-Neighbor Interpolation

One straightforward way to reduce resolution is by specifying a scaling factor less than 1. As highlighted in the provided reference, you can shrink an image to a percentage of its original size.

The reference specifically mentions:

  • Shrinking an image to 40% of the original size.
  • Using the 'nearest' interpolation method.
  • Stating that 'nearest' is the fastest but offers the lowest quality.
  • Providing the example syntax: J = imresize(I,0.4,'nearest');

Here's how you apply this:

Let I be your original image. To reduce its resolution to 40% of the original dimensions using nearest-neighbor interpolation, you use the command:

J = imresize(I, 0.4, 'nearest');
  • I: The original image matrix.
  • 0.4: The scaling factor. This means the output image will have 40% of the original number of rows and 40% of the original number of columns.
  • 'nearest': The interpolation method. Nearest-neighbor interpolation selects the value of the nearest original pixel for the new pixel.

Nearest-neighbor interpolation is the fastest method because it doesn't perform any calculations between pixels. However, it can introduce jagged edges or blocky artifacts, especially at significant downsampling factors, leading to lower visual quality compared to other methods.

After resizing, J will contain the new, lower-resolution image. You can then display both the original and resized images to compare them:

figure;
subplot(1, 2, 1);
imshow(I);
title('Original Image');

subplot(1, 2, 2);
imshow(J);
title('Resized Image (40%, Nearest)');

Other Interpolation Methods

While the reference focuses on 'nearest', imresize supports other interpolation methods for different quality and performance trade-offs. These are typically used for upsampling but can also influence the result during downsampling:

  • 'bilinear': Interpolates using the values of the four nearest pixels. Generally smoother than 'nearest'.
  • 'bicubic': Interpolates using the values of the sixteen nearest pixels. Typically produces the smoothest results but is the slowest.

Choosing the right interpolation method depends on your specific needs regarding processing speed versus image quality.

Reducing Resolution to a Specific Size

Instead of a factor, you can specify the exact number of rows and columns for the output image. For example, to resize an image I to be 200 rows by 300 columns:

J_specific_size = imresize(I, [200 300]); % Uses default interpolation (often bicubic)

You can still specify the interpolation method:

J_specific_size_nearest = imresize(I, [200 300], 'nearest');

Summary Table

Here's a quick overview of common imresize usage for reducing resolution:

Method Syntax Description Interpolation Example Speed Quality
Scale Factor (e.g., 40%) J = imresize(I, 0.4, 'nearest'); Shrink by a percentage. Factor < 1. 'nearest' Fastest Lowest
Scale Factor (e.g., 50%) J = imresize(I, 0.5, 'bilinear'); Shrink by a percentage. Factor < 1. 'bilinear' Medium Medium
Scale Factor (e.g., 75%) J = imresize(I, 0.75, 'bicubic'); Shrink by a percentage. Factor < 1. 'bicubic' Slowest Highest
Specific Size (e.g., 200x300) J = imresize(I, [200 300], 'nearest'); Shrink to exact number of rows and columns. 'nearest' Fastest Lowest

By using the imresize function with a scaling factor less than 1 or specifying a smaller target size, you can effectively reduce the resolution of your image in Matlab. Remember to choose an interpolation method that balances your requirements for speed and visual quality.

Related Articles