The machines are running, the production line is pressing for progress, yet the display screen has already thrown a tantrum.

Bright areas shine so brightly they appear almost white, while dark areas recede into shadow; the defect’s characteristics are simply hard to look at.

The engineer stared at the parameters and tweaked them for a long time, but the real question to ask first might not be, “Can the algorithm be tuned any further?”—but rather—Has this beam of light been aimed correctly after all?

This is a very typical—and often underestimated—issue in machine vision projects.

The camera is responsible for “seeing,” while the algorithm is responsible for “judging.”

But before them, the light source determines:

What on earth did the camera see?

Poor lighting can lead to overexposure, deep shadows, and uneven illumination, potentially resulting in the loss of critical image information.

No matter how hard the algorithm tries afterward, it can only keep working on an image with inherently suboptimal conditions.

Conversely, when the light source is chosen appropriately, target features become clearer and contrast improves, enabling more efficient subsequent algorithmic processing and, consequently, a system that is naturally easier to stabilize.

So, the biggest fear for a visual project isn’t difficulty—it’s going in the wrong direction.

01｜The display is off—don’t rush to wrestle with the algorithm just yet.

In machine vision inspection, the first step is essentially to obtain a “usable image.”

And when it comes to image quality, the light source is one of the key factors.

Different light sources illuminate an object’s surface in distinct ways, producing entirely different light-and-shadow effects.

Ultimately, what matters is the target’s visibility, edges, contrast, and whether its fine details can be reliably “seen” by the camera.

The common pitfalls on site generally fall into these three categories.

First, overexposure.

The light is too strong—the image is washed out and details are completely lost.

The feature you want to detect hasn’t yet been incorporated into the algorithm and may have already been lost during the imaging stage.

Second, shadow interference.

If the illumination angle is inappropriate, shadows may appear on the object’s surface or at its edges, which can bias edge detection.

Third, uneven lighting.

The same scene—bright here, dark there.

The greater the luminance disparity across different regions, the more challenging it becomes to stably select image features.

So, a light source isn’t just about “making things bright.”

The real goal is:

Highlight the features that should be visible, and suppress the interference that shouldn’t be there.

02｜10 Common Light Sources: Though They All Emit Light, Their Characteristics Are Completely Different

The most common pitfall when selecting a light source is thinking that:

“Anyway, they’re all LEDs—pretty much the same.”

Once you actually get the project up and running, you’ll find—

It’s quite a bit off.

1. Annular light source: Light is positioned around the subject; different angles yield different effects.

A ring light arranges LED beads in a circular pattern, directing light around the object and highlighting its three-dimensional features.

Among them, two approaches are particularly common.

High-angle ring light: When the beam’s angle with the vertical is less than 45 degrees, it is better suited for detecting external contours, such as the outer edges of keyboard keys.

Low-angle ring light: When the beam’s angle with the vertical exceeds 45 degrees, it is better suited for highlighting surface scratches, minute irregularities, and associated character features.

In addition, there is alsoRing-shaped shadowless light source.

Its features include more uniform illumination, greater compatibility with highly reflective objects, and reduced interference from reflected light.

2. Linear light source: Flexible angle, specifically “extracting features” from the side

A linear light source arranges LED beads in a long strip, enabling illumination of the target from one or multiple sides and emphasizing edge features.

It is commonly used in applications such as metal surface inspection, character reading, and LCD panel testing.

One of its major advantages is its high degree of flexibility:

The irradiation angle and installation distance can both be flexibly adjusted to meet detection requirements.

3. Bottom backlight: It doesn’t get bogged down by surface details and instead brings out the contours directly.

The bottom-mounted backlight arranges LED beads into a uniform light panel, typically positioned at the base of an object, making it ideal for large-area illumination.

Its advantages are very straightforward:

Enhance the object’s contour and size information.

Therefore, this type of light source can be used for detecting transparent objects, measuring the dimensions of metal parts, and inspecting the edge seams of plastic films.

4. Center-hole backlight: A hole is cut in the middle, allowing the camera to “see” through it.

The mid‑hole backlight can be regarded as a modified design of the bottom‑type backlight, with an opening left in the center.

The light source is positioned directly above the object, and the camera captures the image through the central aperture.

This structure enables more precise acquisition of surface reflection information, making it suitable for high-precision surface quality inspection.

5. Coaxial light source: Align the lighting direction with the camera’s line of sight along a single axis.

The coaxial light source emits light from the side; after passing through a half‑transmissive, half‑reflective mirror, the light aligns with the camera lens axis.

The advantage of this approach is that it delivers more uniform, brighter illumination, making it particularly well-suited for quality inspection of smooth surfaces.

In practical applications, a coaxial light source can also be used in conjunction with a backlit source featuring a central aperture to eliminate shadows in the central region and enhance image detail.

6. Spherical integrated light source (bowl light): When encountering highly reflective surfaces, uniform illumination is crucial.

The spherical integral light source employs bottom-mounted LED beads and a hemispherical reflective coating to achieve diffused illumination.

Its most distinctive feature is:

The lighting is very even.

For highly reflective objects such as metals and glass, this lighting method is particularly valuable.

Common applications include dashboard gauges and metal can coding inspection, among others.

7. Line Light Source: A Common Companion to Line-Scan Cameras

The linear light source consists of high-brightness LEDs and, after being collimated by a light-guiding column, produces a bright luminous strip.

It is typically used in conjunction with line-scan cameras and is well suited for surface defect inspection.

For example, surface defects on LCD screens, glass scratches, and textile imperfections are all common application areas for it.

8. Point light source: Limited in range, but with a very tight beam.

Point light sources consist of small, high-brightness LEDs and are typically used in conjunction with telecentric lenses to provide compact, high-intensity illumination.

Its focusing performance is outstanding, making it ideal for applications such as scratch detection on glass surfaces and substrate alignment in liquid crystal displays.

9. Infrared Light Source: When Surface Interference Is Severe, Try a Different Perspective

Infrared light sources are suitable for night-vision applications and temperature‑change detection.

Because infrared light sources have strong penetrative power, they can pass through the product’s surface, thereby reducing interference caused by surface illumination.

It is widely used in areas such as heat-seal testing and hot-melt adhesive testing.

10. Ultraviolet light source: Some features are barely visible under ordinary lighting, but become distinct when illuminated with ultraviolet light.

Ultraviolet light sources can induce fluorescence.

Many phosphors exhibit distinct color changes under ultraviolet irradiation, making UV light sources indispensable in applications such as banknote authentication and photolithographic exposure.

As you can see, selecting a light source isn’t simply a matter of “using whichever one is brightest.”

Different kinds of light, at their core, serve to “extract” the target’s features from the background and interference in distinct ways.

03｜How should you choose a light source? Don’t start by asking about the model—first consider these four factors.

Now that we know the type of light source, the next step is where the real headaches begin on the project site:

Which one should I choose after all?

Don’t rush to focus on brightness and installation location right away.

First, let’s clarify the following four questions.

First, examine the material and surface characteristics.

Metals, glass, and plastics exhibit different light‑reflecting properties.

Especially when dealing with reflective surfaces, improper lighting can easily lead to reflection-induced interference on site.

So the first step is to figure out how the object being measured reflects light.

Second, consider the size and complexity of the target being inspected.

Large-area targets and fine, small targets clearly have different requirements for the light source.

How big is the object?

What are the surface features?

What are the requirements for detection accuracy?

All of these factors will influence the choice of light source.

Third, examine the lighting angle and the reflection effect.

Often, even when the light type is correct, an incorrect angle can still produce the wrong effect.

Different illumination angles can alter the way surface features are perceived.

Therefore, the illumination angle should be determined in conjunction with the object’s geometry and the type of defect.

Fourth, consider whether the application scenario has any special requirements.

For example, when dealing with highly reflective objects, you might consider using a ring-shaped, shadowless light source.

If you need to detect surface defects on an LCD panel, a linear light source may be more suitable.

That’s also why the visual‑project team won’t stick around to patiently “tune in the dark” with you.

First, clarify the material, target, angle, and scene; then discuss light source selection—this will make your approach much more efficient.

04｜Truly stable vision systems often begin with “directing the light correctly.”

In machine vision, lighting is not an optional add-on, nor is it a last‑minute component tacked on at the end of a project.

It directly affects image quality and, in turn, impacts subsequent inspection results.

Overexposure causes loss of detail.

The shadow is incorrect, and the edges may be wrong.

Uneven illumination makes features difficult to stabilize.

Moreover, when the light source is matched to the material, dimensions, surface characteristics, inspection target, and illumination angle of the object being inspected, the image becomes sharper and the features more distinct.

Subsequent algorithms can also operate more efficiently, naturally ensuring greater system stability.