The optical foundation of machine vision: the conversion from light to signal

Introduction to Machine Vision

The fundamental task of a machine vision system is to acquire a clear, distortion-free image of the object being measured. This process begins with the illumination of the object by a light source and ends with the camera sensor converting the light signal into an electrical signal. In this chain, the optical lens undertakes two core functions: light collection and imaging.

Machine Vision: Several Factors Affecting Image Quality

From a geometrical optics perspective, the lens needs to converge the diffused light emitted from the object along a specific optical path and form a real image on the sensor surface of the camera. The imaging quality of this process is determined by several optical parameters, the most critical of which include:

Resolution: The ability of a lens to resolve object details, usually measured in line pairs per millimeter (lp/mm). It must match the pixel size of the camera to fully utilize the sensor’s potential.

Distortion: This refers to the image distortion caused by changes in lens magnification with the field of view. In traditional lenses, distortion typically manifests as barrel or pincushion distortion, causing the size of objects at image edges to deviate from reality, a major source of error in precision measurements.

Depth of field: The object-space depth that allows for a sharp image on the image plane. When detecting three-dimensional objects with height differences, a large depth of field is essential to ensure the sharpness of the entire object’s outline.

In ordinary non-telecentric lenses, light enters the lens at a certain angle, meaning the distance of the object from the lens (object distance) directly affects the image size. For applications requiring high-precision dimensional measurements, this perspective error caused by changes in object distance or the object’s thickness is a systematic bias that cannot be completely eliminated by post-processing algorithms.

The core principle of telecentric lenses: Parallel light paths and parallax elimination

The starting point of telecentric lens design is to completely avoid the parallax effect of traditional lenses. Its core optical principle lies in controlling the path of the principal ray by placing the aperture stop at the focal plane of the lens system.

We can understand the underlying physical mechanism of this lens through the following three structural designs:

1. Object-Side Telecentric Lens

By placing an aperture stop on the image-side focal plane of the lens, only the principal rays parallel to the optical axis (i.e., the central rays from various points on the object side) can pass through the aperture stop and participate in imaging. This means that light reflected from different distances from the object, as long as its direction is parallel to the optical axis, will ultimately have a constant image height on the image plane. In short, the magnification no longer changes with the object distance, fundamentally eliminating measurement errors caused by inaccurate object placement or surface undulations.

2. Image-Side Telecentric Lens

The aperture stop is placed on the object-side focal plane of the lens. In this case, after passing through the aperture stop, the principal rays in image-side space are parallel to the optical axis. The advantage of this design is that even if there is a slight positional error (image distance change) in the camera sensor during installation, the image size will not change, ensuring imaging stability.

3. Double Telecentric Lens

As the name suggests, it combines the advantages of the two structures mentioned above, achieving parallel optical paths on both the object and image sides. This is currently the most powerful lens structure in the field of precision measurement in machine vision. It not only eliminates errors caused by changes in object distance but also eliminates the influence of changes in image distance, achieving constant magnification throughout the entire working range. According to optical design specifications, modern high-performance double telecentric lenses can control optical distortion to below 0.07%, or even lower, greatly restoring the true geometric shape of objects.

It is this unique parallel optical path design that gives telecentric lenses near-perfect telecentricity. Although completely distortion-free lenses do not exist in engineering practice, through precise optical design and component manufacturing, the divergence angle of the principal ray can be controlled within an extremely small range (e.g., within 0.06°), thus meeting the measurement requirements at the micrometer level.

The Role of MOK Optics Optical Components in Machine Vision

To achieve the aforementioned complex parallel optical path and control aberrations, the support of high-quality optical components is indispensable. MOK Optics’ deep expertise in optical cold processing and coating provides complete solutions for machine vision systems, from core imaging to beam control.

1. High-Precision Spherical and Aspherical Lenses

Telecentric lenses typically consist of multiple refractive lenses (e.g., some designs require up to 10 lenses) to correct various monochromatic aberrations, including spherical aberration, coma, and field curvature. MOK Optics’ precision spherical lenses, through rigorous optical design, enable precise control of the optical path. In particular, their perforated spherical lenses, designed for specific needs, can achieve complex beam shaping within limited space, which is crucial for inspection scenarios requiring selective light transmission or enhanced image contrast.

2. Optimized Illumination Systems

In machine vision, illumination determines the raw quality of the image. Telecentric lenses are typically used in conjunction with telecentric illumination to achieve optimal edge detection results. MOK Optics offers a variety of lenses that play a crucial role in illumination systems:

Powell Lenses: In 3D inspection or laser line scanning applications requiring structured light, Powell lenses can shape a Gaussian-distributed laser beam into a straight spot with uniform energy. This uniform illumination eliminates detection errors caused by excessive brightness at the laser center and insufficient brightness at the edges, ensuring a stable and clear contour line on the object surface. This is particularly important for 3D dimensional measurement and surface scratch detection.

Beam Shaping and Homogenization: In backlighting or coaxial illumination, microlens arrays or specially designed homogenizing lenses can transform point or line light sources into uniformly bright surface light sources, ensuring consistent illumination across the entire field of view, thereby highlighting the edge features of the object being measured.

3. Applications of Coating Technology

Surface reflections from optical components cause light energy loss and stray light, severely affecting image contrast. MOK Optics employs various precision coating technologies on its lenses:

Antireflective coating: By depositing multiple layers of dielectric film on the lens surface, reflectivity in specific wavelengths (such as visible and near-infrared) can be effectively reduced, increasing light transmittance and allowing more object information to reach the sensor.

Filter coating: In certain applications, such as detection using laser illumination of specific wavelengths, bandpass filters need to be integrated into the lens. MOK Optics can directly deposit wavelength-selective films on the lens substrate according to customer needs, precisely filtering signal light and shielding against ambient light interference.

Conclusion

As the entry point for optical information, the quality of the lens directly determines the upper limit of the entire system’s performance. Telecentric lenses, with their unique parallel optical path design, fundamentally solve the core problem of perspective error that plagues precision measurement. MOK Optics provides a solid optical foundation for telecentric imaging systems and machine vision illumination by manufacturing high-precision spherical lenses and Powell lenses and applying advanced optical coating technologies. In the future, as the precision requirements in display technology, semiconductor packaging, and new energy manufacturing continue to increase, the role of optical components in machine vision will become increasingly crucial. Starting from optical principles, we continuously optimize component performance to help intelligent manufacturing develop in a more precise, faster, and more stable direction.