extractpatches

Extract patches from image at specified ROIs

Since R2026a

collapse all in page

Syntax

patches = extractpatches(I,bboxes)

Description

Add-On Required: This feature requires the Automated Visual Inspection Library for Computer Vision Toolbox add-on.

patches = extractpatches(I,bboxes) extracts patches from image I at the ROI locations specified by the bounding boxes bboxes.

Note

This functionality requires the Automated Visual Inspection Library for Computer Vision Toolbox™. You can install the Automated Visual Inspection Library for Computer Vision Toolbox from Add-On Explorer. For more information about installing add-ons, see Get and Manage Add-Ons.

example

Examples

collapse all

This example uses:

Open Live Script

Load an image containing objects to count into the workspace.

I = imread("coins.png");

Display the image using the imageshow function.

hImage = imageshow(I);

To interactively define bounding box ROIs on the image, use the uiselectboxes function. Click a start point in the image and drag boxes around the desired objects. Click Accept in the top-right corner of the figure window when you are satisfied with your selections.

bboxes = uiselectboxes(hImage)
bboxes = 4×4

    195.4484     28.4718    69.5974    63.5455
    321.0263    101.0952    55.9805    58.2500
    224.9516    116.2251    65.0584    61.2760
    165.1886     85.9653    55.2240    52.1980

This GIF shows how to interactively select bounding boxes by creating and dragging the boxes to different positions on the image.

Extract the patch tiles from the image using the extractpatches function.

tiles = extractpatches(I,bboxes)
tiles=4×1 cell array
    64×70 uint8
    58×56 uint8
    62×65 uint8
    52×55 uint8

Display the bounding boxes defining the image patches overlaid on the image using the insertShape function.

annotatedImage = insertShape(I,"rectangle",bboxes,"LineWidth",3);
imshow(annotatedImage)

Input Arguments

collapse all

Input image from which to extract patches, specified as one of these options:

Image TypeData Format
Grayscale H-by-W numeric matrix
RGB H-by-W-by-3 numeric array

Axis-aligned rectangular bounding box ROIs that enclose patches, specified as an M-by-4 numeric matrix. M is the number of bounding box ROIs. Each row of the matrix represents a bounding box, and is of the form [x y w h], where:

  • x and y specify the upper-left corner of the rectangle.

  • w specifies the width of the rectangle, which is its length along the x-axis.

  • h specifies the height of the rectangle, which is its length along the y-axis.

You can interactively select bounding box ROIs for an image using the uiselectboxes function.

Output Arguments

collapse all

Image patches, returned as an M-by-1 cell array. M is the number of bounding box ROIs specified using the bboxes name-value argument. Each element of the array is an image patch, returned as one of these values depending on the format of the input image I.

  • RGB — Each patch is a P-by-Q-by-3 numeric array.

  • Grayscale — Each patch is a P-by-Q numeric matrix.

P and Q are the height and width dimensions of each patch, respectively.

Version History

Introduced in R2026a

See Also

|