Student Competition: Computer Vision Training, Automating Image Registration

Learn to estimate geometric transforms, apply the transform, and specify image coordinates using panoramic stitching as an example.

You’ll start by performing feature matching between two consecutive images and estimating the geometric transform between two images. The two images will be superimposed onto a larger image to create a super image that encompasses both images. This workflow will be demonstrated on a video of a panning vehicle to create a panoramic image.

Besides learning how to perform image stitching, learning how to estimate geometric transformations enables the user to perform accurate object detection and estimate the motion of an object within a scene.

We’d like to thank the Cedarville University RoboBoat team for providing a video from their vehicle.

Additional Information:

• Geometric Transformations – Functions, Concepts, and Examples
• estimateGeometricTransform function ‘transfor mType’ property documentation
• rng function documentation
• imref2d function documentation
• outputLimits function documentation
• vision.AlphaBlender System object documentation