The Simplest Professional Evaluation Method for an Industrial Lens

An industrial lens differs from other types of lenses (such as DSLR lenses, dashcam lenses, and smartphone lenses) in terms of usage, performance, design, and manufacturing. So, what parameters should we measure to evaluate the performance of an industrial lens?

Resolution

Resolution is the most critical parameter for an industrial lens, as it defines how much detail the lens can capture. Essentially, a lens acts as a low-pass filter, meaning high-frequency details are lost during imaging.

To test lens resolution, a resolution chart is required. The USAF1951 chart is commonly used to measure the smallest resolvable line width, providing an accurate reading of the object-side resolution. This value is influenced by the working distance and is best converted to image-side resolution for comparison.

During testing, adjust the center of the field of view to achieve the sharpest focus, then measure the resolution at the center and four corners. It is essential to maintain the same aperture and working distance for all lens comparisons. High-quality lenses will have minimal image quality differences between the center and corners. For an industrial lens, the center resolution will generally be better than the edges.

Distortion

Distortion refers to the inaccuracy of the magnification ratio, where object-space coordinates do not linearly correspond to image-space coordinates. Distortion tends to increase as you move further from the center of the field of view.

However, optical distortion can be difficult to measure directly, so a more practical approach is to calculate TV distortion using the formula: TV Distortion = ΔH / H. Where ΔH is the difference between the actual and theoretical positions of the image points.

When software like Halcon is used for distortion calibration, it typically relies on optical distortion. However, for distortion measurement, most software uses TV distortion. For the same industrial lens, the value of optical distortion is generally greater than TV distortion. Using software like Imatest is the fastest and most accurate way to measure distortion. However, if purchasing Imatest is not an option, you can write a program to calculate distortion based on the definition of optical distortion. The process involves reading the object-space coordinates, finding the deviation from the theoretical coordinates, and dividing it by the distance to the center point to determine the distortion value at that point. Since optical distortion is nonlinear, particularly with mixed distortion, effects like pincushion distortion in the center and barrel distortion at the edges may occur. To accurately represent this, it is essential to collect as many data points as possible and fit a distortion curve.

Relative Illumination

Relative illumination refers to the difference in brightness between the center and the edges of an image, which is crucial for reliable edge detection. Excessive brightness variation can lead to inconsistent feature extraction.

To test relative illumination, use a large, uniformly lit light source. First, ensure the uniformity of the light source. Capture an image, measure the average grayscale values of the center and edge regions, and apply the formula: Relative Illumination = (Center Grayscale - Edge Grayscale) / Center Grayscale. Since brightness generally decreases from the center to the edges, measurements from one corner are typically sufficient. Note that relative illumination varies with working distance. Shorter distances and smaller apertures generally result in higher relative illumination and more uniform brightness.

Contrast

Contrast is another crucial parameter often confused with resolution. While resolution corresponds to the high-frequency region of the Modulation Transfer Function (MTF) curve, contrast corresponds to the low-frequency region. To test contrast, capture an image of a resolution chart with bold line patterns. Draw a line across the edge of a dark-to-light transition and analyze the continuous grayscale values. Then calculate the contrast. In applications such as smartphone panel inspection, contrast is often more critical than resolution because defect points and normal areas may only differ by a few grayscale values. A low-contrast lens will struggle to identify these defects.

Stray Light

Compared to security lenses, DSLR lenses, and automotive lenses, industrial lens has lower stray light requirements. This is because most machine vision applications involve even lighting in a controlled environment, reducing the likelihood of direct light entering the lens. Stray light issues are complex and vary depending on environmental lighting conditions. To perform a simple stray light test, use a strong flashlight to observe how the lens handles direct and indirect light sources. If stray light becomes an issue during a project, consider adjusting the light source angle or using a polarizer to mitigate the effect.