Resize image


B = imresize(A, scale)
gpuarrayB = imresize(gpuarrayA,scale)
B = imresize(A, [numrows numcols])
[Y newmap] = imresize(X, map, scale)
[...] = imresize(...,method)
[...] = imresize(..., parameter, value, ...)


B = imresize(A, scale) returns image B that is scale times the size of A. The input image A can be a grayscale, RGB, or binary image. If scale is between 0 and 1.0, B is smaller than A. If scale is greater than 1.0, B is larger than A. By default, imresize uses bicubic interpolation.

gpuarrayB = imresize(gpuarrayA,scale) performs the resize operation on a GPU. The input image and the output image are gpuArrays. When used with gpuArrays, imresize only supports cubic interpolation and always performs antialiasing. This syntax requires the Parallel Computing Toolbox™.

B = imresize(A, [numrows numcols]) returns image B that has the number of rows and columns specified by [numrows numcols]. Either numrows or numcols may be NaN, in which case imresize computes the number of rows or columns automatically to preserve the image aspect ratio.

[Y newmap] = imresize(X, map, scale) resizes the indexed image X. scale can either be a numeric scale factor or a vector that specifies the size of the output image ([numrows numcols]). By default, imresize returns a new, optimized colormap (newmap) with the resized image. To return a colormap that is the same as the original colormap, use the 'Colormap' parameter (see below).

[...] = imresize(...,method) specifies the interpolation method used. method can be a text string that specifies a general interpolation method or an interpolation kernel, specified in the following table, or a two-element cell array, of the form {f,w}, that specifies an interpolation kernel, where f is a function handle for a custom interpolation kernel and w is the custom kernel's width.f(x) must be zero outside the interval -w/2 <= x < w/2. Your function handle f may be called with a scalar or a vector input..



Nearest-neighbor interpolation; the output pixel is assigned the value of the pixel that the point falls within. No other pixels are considered.


Bilinear interpolation; the output pixel value is a weighted average of pixels in the nearest 2-by-2 neighborhood


Bicubic interpolation (the default); the output pixel value is a weighted average of pixels in the nearest 4-by-4 neighborhood

'box'Box-shaped kernel
'triangle'Triangular kernel (equivalent to 'bilinear')
'cubic'Cubic kernel (equivalent to 'bicubic')
'lanczos2'Lanczos-2 kernel
'lanczos3'Lanczos-3 kernel

[...] = imresize(..., parameter, value, ...) you can control various aspects of the resizing operation by specifying parameter/value pairs with any of the previous syntaxes. The following table lists these parameters.

'Antialiasing'A Boolean value that specifies whether to perform antialiasing when shrinking an image. The default value depends on the interpolation method. If the method is nearest-neighbor ('nearest'), the default is false; for all other interpolation methods, the default is true.
'Colormap'A text string that specifies whether imresize returns an optimized colormap or the original colormap (Indexed images only). If set to 'original', the output colormap (newmap) is the same as the input colormap (map). If set to 'optimized', imresize returns a new optimized colormap. The default value is 'optimized'.
'Dither'A Boolean value that specifies whether to perform color dithering (Indexed images only). The default value is true.
'Method'As described above
'OutputSize'A two-element vector, [numrows numcols], that specifies the size of the output image. If you specify NaN for one of the values, imresize computes the value of the dimension to preserve the aspect ratio of the original image.
'Scale'A scalar or two-element vector that specifies the resize scale factors. If you specify a scalar, imresize uses the value as the scale factor for each dimension. If you specify a vector, imresize uses the individual values as the scale factors for the row and column dimensions, respectively.

Class Support

The input image can be numeric or logical and it must be nonsparse. The output image is of the same class as the input image. An input image that is an indexed image can be uint8, uint16, or double.

The input image gpuarrayA can be can be a single- or double-precision gpuArray. The output gpuArray image is of the same underlying class as the input image.


Shrink image by factor of two, using default interpolation and antialiasing.

I = imread('rice.png');
J = imresize(I, 0.5);
figure, imshow(I), figure, imshow(J)

Shrink image by factor of two, performing the operation on a GPU.

I = im2double(gpuArray(imread('rice.png')));
J = imresize(I, 0.5);
figure, imshow(I), figure, imshow(J)

Shrink by factor of two using nearest-neighbor interpolation. This is the fastest method, but it has the lowest quality.

J2 = imresize(I, 0.5, 'nearest');

Resize an indexed image

[X, map] = imread('trees.tif');
[Y, newmap] = imresize(X, map, 0.5);
imshow(Y, newmap)

Resize an RGB image to have 64 rows. Let imresize calculate the number of columns necessary to preserve the aspect ratio.

RGB = imread('peppers.png');
RGB2 = imresize(RGB, [64 NaN]);

Resize an RGB image, performing the operation on a GPU.

RGB = gpuArray(im2single(imread('peppers.png')));
RGB2 = imresize(RGB, 2);

More About

collapse all


The function imresize changed in version 5.4 (R2007a). Previous versions of the Image Processing Toolbox™ used a somewhat different algorithm by default. If you need the same results produced by the previous implementation, use the function imresize_old.

For bicubic interpolation, the output image may have some values slightly outside the range of pixel values in the input image. This may also occur for user-specified interpolation kernels.

Was this topic helpful?