Is it possible to use undistortImage using a custom image I have?

Hi, I have acquired the image of a checkerboard (see file w_teleconverter.mat), to visualize it, scale it between [0 10000]. This is basically to assess if I can perform image distortion correction on the image that I have. Note that the image is the sharpest possible, and the camera is mounted to a spectrometer, so only part of the checkerboard is seen. However the examples that I see online are for standard calibration patterns. Would it be possible to use undistortImage for this image I have?

[Image Analyst]

Can you insert the image right here so we don't have to write a MATLAB program just to see your image?

[Akshay Arvind]

Yeah, sure. Here it is:

Answer 1

[colordepth]

MATLAB's "undistortImage" function requires accurate camera intrinsic parameters (https://www.mathworks.com/help/vision/ref/undistortimage.html#btyx3m1-3). Estimating these parameters from a single, incomplete checkerboard image is difficult.

Your checkerboard lacks a complete view, with only one column fully visible and parts of others occluded. MATLAB's documentation (https://www.mathworks.com/help/vision/ug/prepare-camera-and-capture-images-for-camera-calibration.html#:~:text=white%20border%20surrounding) stresses the need for a white border around the completely visible pattern for accurate calibration.

The current image will result in inaccurate camera intrinsics and an incorrect "undistortImage" output. I suggest printing a smaller checkerboard and capturing at least three different images to improve calibration. Follow this guide on single camera calibration (https://www.mathworks.com/help/vision/ug/using-the-single-camera-calibrator-app.html) for detailed steps, including how to prepare the checkerboard images.

[Akshay Arvind]

Hi, thanks for the answer! Obviously part of the image is occluded because the camera is mounted to a spectrometer, but I will remember to print an even smaller checkerboard with a white background.