insertObjectInImage
R2026bSyntax
Description
[
inserts an object from a source image, newImage,boundingBox,newMask] = insertObjectInImage(destinationImage,sourceImage,mask)sourceImage, into a destination
image, destinationImage. By default, the
insertObjectInImage function inserts an object from the source image into a
randomized location on the destination image. The function returns the result as an
augmented image, newImage, a bounding box,
boundingBox, and a corresponding object mask,
newMask.
[
specifies options using one or more name-value arguments, in addition to the input arguments
from the previous syntax. For example,
newImage,boundingBox,newMask] = insertObjectInImage(___,Name=Value)BoundaryConstraintMode="inbounds-exact" specifies to insert objects
only when they are completely within the destination image.
Examples
Read the destination image into the workspace and display the image.
destImage = imread("parkroad.jpg");
imshow(destImage)
Read the source image containing the object, the small dog, into the workspace and display the image.
sourceImage = imread("DogTrio.jpg");
imshow(sourceImage)
Load the binary object mask into the workspace.
load("dogMask.mat");Insert the object from the source image into a randomized location on the destination image. Generate the augmented synthetic image, the object mask, and the corresponding bounding box.
[newImage,boundingBox,newMask] = insertObjectInImage(destImage,sourceImage,dogMask);
Annotate the image at the position specified by the generated bounding boxes using the insertObjectAnnotation function.
annotedNewImage = insertObjectAnnotation(newImage,"rectangle",boundingBox,"Dog");
Display the annotated synthetic image.
figure
imshow(annotedNewImage)
title("Annotated Synthetic Image")
Insert an anomaly defect into a normal image of a metal plate using industrial part images from the RobustAD data set. Use this technique to generate synthetic training data for anomaly detection models.
Load Destination Image
Load a normal, defect-free image of a metal plate into the workspace, and display the image. This is the destination image into which you insert the anomaly.
destImage = imread("metalplate.jpg"); imshow(destImage) title("Normal Plate Image (Destination)")
Load Anomalous Source Image
Load an anomalous image into the workspace, and display the image. Extract an anomaly object from this source image.
sourceImage = imread("metalplate_anomaly.jpg"); imshow(sourceImage) title("Anomalous Image (Source)")
Load Anomaly Mask
Load the binary mask that defines the boundary of the anomaly in the source image into the workspace, and display the mask. The mask is a logical matrix where true pixels indicate the anomaly region.
anomalyMask = imread("metalplate_anomalymask.png"); imshow(anomalyMask) title("Anomaly Mask")
Load Region Constraint Mask
Load the normal mask corresponding to the destination image into the workspace, and display the mask. This mask defines the valid region on the plate where you can insert defects, ensuring you place the anomaly object only on the plate surface.
regionMask = imread("metalplate_normalmask.png"); regionMask = logical(regionMask); imshow(regionMask) title("Region Constraint Mask (Valid Insertion Area)")
Insert Anomaly into Normal Image
Use the insertObjectInImage function to insert the anomaly from the anomalous source image into the normal destination image. Specify the RegionConstraintMask name-value argument to restrict placement to the valid plate surface area. The function returns the augmented image, the bounding box of the inserted object, and a new mask indicating the object location in the output image.
[newImage,boundingBox,newMask] = insertObjectInImage(destImage,sourceImage,anomalyMask, ...
RegionConstraintMask=regionMask);Display Synthetic Anomaly Image
Annotate the synthetic image with a bounding box around the inserted anomaly by using the insertObjectAnnotation function. Display the annotated synthetic image.
annotatedImage = insertObjectAnnotation(newImage,"rectangle",boundingBox,"Anomaly"); imshow(annotatedImage) title("Annotated Synthetic Anomaly Image")
Visualize New Mask
Visualize the returned mask, which corresponds to the location of the inserted anomaly in the destination image. You can use this mask as ground truth when training segmentation models.
imshow(newMask)
title("Inserted Anomaly Mask in Destination Image")
Apply Poisson Blending
To create more realistic anomalies, specify the BlendMethod name-value argument as "poisson" to seamlessly blend the anomaly into the destination image while constraining placement to the valid region.
[blendedImage,boundingBox2,blendedMask] = insertObjectInImage(destImage,sourceImage,anomalyMask, ... BlendMethod="poisson",RegionConstraintMask=regionMask); imshow(blendedImage) title("Synthetic Image with Poisson Blending")
Copyright 2026 The MathWorks, Inc.
Input Arguments
Name-Value Arguments
Output Arguments
References
[1] He, Kaiming, Jian Sun, and Xiaoou Tang. “Guided Image Filtering.” IEEE Transactions on Pattern Analysis and Machine Intelligence 35, no. 6 (June 2013): 1397–1409. doi:10.1109/TPAMI.2012.213.
[2] Pérez, Patrick, Michel Gangnet, and Andrew Blake. “Poisson Image Editing.” In ACM SIGGRAPH 2003 Papers, 313–18. San Diego California: ACM, 2003. doi:10.1145/1201775.882269.
