Do you ever face this dilemma? When taking a photo, a perfectly straight wall appears curved like a wave; a square building looks distorted in the image? Don't worry—it's not a faulty camera or a problem with Earth's gravity, but rather lens distortion at work!

What is lens distortion?

Lens distortion, as the name suggests, refers to image deformation caused by the optical structure of a lens. Simply put, objects that should appear straight may look distorted when photographed—similar to the appearance of Transformers. While this phenomenon may occur in everyday photography, its impact in industrial machine vision is far more significant. Machine vision is not merely for imaging; it also handles precise measurement and recognition tasks. Distortion can lead to data errors and even disrupt the normal operation of the entire system.

Two major types of distortion: radial distortion and tangential distortion

In industrial vision, there are two common types of lens distortion that directly affect the accuracy of machine vision systems. The following provides a detailed analysis of each type:

Radial Distortion

The most common type of distortion. Due to lens design or manufacturing processes, edge distortion in images is typically more pronounced than central distortion. Common manifestations include:

Barrel distortion: The image bulges outward, causing straight lines to curve outward, resembling a barrel.

Pincushion distortion: The image appears concave inward, with straight lines curving inward and the four corners sunken like a pillow.

Tangential Distortion

This distortion is typically caused by an incorrect lens mounting angle. The lens is not fully parallel to the image sensor, resulting in distortion at both the center and edges of the image.

Why is human vision less susceptible to distortion?

You might wonder why distorted photos taken in daily life don't seem particularly problematic. In fact, the human brain possesses an inherent self-correction mechanism. When distortion is less than 2%, it is generally imperceptible to us. However, in industrial settings where precision requirements are extremely high, even minor distortions can lead to measurement errors or recognition failures.

How should distortion be addressed? Correction is the key!

Given the significant impact of distortion, how should we address it? The answer is—correction!

Software Correction

For radial distortion, software correction is the most common and straightforward method. By using the grid calibration tool, the software automatically calculates the distortion coefficient and corrects the image to restore its normal shape.

This correction method is simple and easy to implement, and nearly all machine vision software available on the market already integrates this functionality.

Hardware Correction

Tangential distortion is typically caused by improper lens installation. By ensuring the lens remains parallel to the image sensor during installation, distortion can be significantly reduced.

Camera Calibration: A Essential Lesson in Machine Vision

In industrial machine vision, camera calibration is an indispensable step that ensures the camera can accurately convert "distorted image coordinates" into "real-world coordinates." Through calibration, machine vision systems can perform measurements and positioning operations with greater precision. The common calibration tool is a grid plate, which automatically assists the system in completing calibration and distortion correction.

In summary!

Have you encountered lens distortion issues when using a machine vision system? Share your experience!