Estimate Distortion from a Pattern
Straight board lines bow near the edge of a wide lens.
Estimate Distortion from a Pattern explained visually
Straight board lines bow near the edge of a wide lens. English captions are optional and start off. The complete transcript follows the written lesson.
What you will learn
Straight board lines bow near the edge of a wide lens.
A practical question
In the parking image, a straight painted line bows near the border. The camera lens may be bending its projected position. A calibration pattern gives us many known points to compare with an ideal camera model.
The useful idea
Estimate internal parameters and distortion coefficients from varied pattern views, then use the model to predict where each observed corner should appear. The distortion describes systematic departures from the ideal projection, often more visible away from the image center. It should be estimated from data, not copied from a different lens because the cameras look alike.
Work through the example
Undistort a test photo and inspect straight painted lines across the frame. Then project known corners and examine their errors. A visually straighter line is encouraging, but cropped or shifted borders after correction also need accounting in the image coordinates.
Limits and next step
Save the original and calibration alongside the corrected image. Next, we'll read the remaining reprojection error without letting one average hide bad views.
Worked example
Compare corner residuals before and after estimating radial distortion.
Quick check: test your understanding
Try it yourself: In the situation described above, explain the method and one reason its output might be misleading.
Check the reasoning
A distortion model maps ideal pinhole coordinates to the real lens's observed positions. Overfitting a flexible model to weak board detections can mislead measurement.
What is the main idea behind estimate distortion from a pattern?
A distortion model maps ideal pinhole coordinates to the real lens's observed positions.
How is it used in the example?
Compare corner residuals before and after estimating radial distortion.
What limitation should be checked?
Overfitting a flexible model to weak board detections can mislead measurement.
Further reading
Programming Computer Vision with Python — Jan Erik Solem.
Original ToolsInVision teaching examples. Editorial approach · Suggest a correction
Accessible referenceLesson 22 transcript155 words · Video transcript · Matches the published narration and captionsOpen transcript
In the parking image, a straight painted line bows near the border. The camera lens may be bending its projected position. A calibration pattern gives us many known points to compare with an ideal camera model.
Estimate internal parameters and distortion coefficients from varied pattern views, then use the model to predict where each observed corner should appear. The distortion describes systematic departures from the ideal projection, often more visible away from the image center. It should be estimated from data, not copied from a different lens because the cameras look alike.
Undistort a test photo and inspect straight painted lines across the frame. Then project known corners and examine their errors. A visually straighter line is encouraging, but cropped or shifted borders after correction also need accounting in the image coordinates.
Save the original and calibration alongside the corrected image. Next, we'll read the remaining reprojection error without letting one average hide bad views.