Access Official Technical PDF

Share Summary:LinkedInEmail

A polarization image sensor adds a useful layer of information to industrial machine vision: it captures polarization characteristics that conventional image sensors do not detect. For inspection tasks involving reflections, transparent materials or subtle surface defects, this information can help make features easier to distinguish. Sony Semiconductor Solutions’ Polarsens sensors combine four-directional on-chip polarizers with industrial imaging functions such as global shutter operation, region-of-interest (ROI) modes and external triggering.

How a polarization image sensor captures additional information

A conventional image sensor records light intensity and, in color models, color information. A polarization image sensor also measures how light is polarized. In the described Sony sensors, four differently oriented polarizers are formed over the pixels. The sensor captures the four directional polarization images in one exposure, and subsequent signal processing can use their relative intensities to calculate the direction and degree of polarization (DoP).

DoP describes the extent to which light is polarized, while polarization direction indicates its orientation. These measurements can provide contrast that is not apparent in a standard intensity image. The resulting information can be processed in real time, subject to the capacity of the downstream signal-processing system.

On-chip polarizer design and image quality

The placement of the polarizer matters. In the described sensor design, polarizers are formed on the chip beneath the on-chip lens layer, closer to the photodiode than in conventional arrangements where the polarizer sits above the lens layer. The shorter distance is intended to improve extinction ratio and reduce dependence on the angle at which light enters the sensor.

Extinction ratio compares sensitivity to light aligned with a polarizer’s transmission axis against sensitivity to light aligned with its extinction axis. A higher ratio indicates greater separation between those responses. The sensors also use an anti-reflection layer to help reduce reflectance and avoid poor flare and ghost characteristics.

Because the polarizers are formed during semiconductor processing, the design supports accurate alignment with the pixels. The source also identifies uniformity, mass productivity and durability as advantages of this manufacturing approach compared with conventional polarization sensors.

Industrial inspection applications

Polarization data can help reveal differences in material or surface response that are difficult to see in a normal image. The supplied examples illustrate several potential inspection uses:

  • Glass inspection: Differences in DoP can help identify scratches, stains, fingerprints and dust on a uniform glass surface.
  • Tablet filling inspection: A difference in polarization response between tablets and their aluminum packaging can help determine whether tablets are present.
  • Distortion inspection: Polarization direction can indicate both the presence and direction of distortion in a plane.
  • Reflection reduction: Polarization information can be used to remove reflections. The four-directional arrangement can address reflections across multiple planes simultaneously.

These examples show how polarization can complement conventional imaging rather than replace it. Engineers can evaluate whether the additional signal helps distinguish the specific defect, material or condition their inspection system needs to detect.

Global shutter and imaging controls for production lines

Industrial inspection often involves moving objects. With a rolling shutter, different image rows are captured at different times, which can cause focal-plane distortion when the subject moves. The IMX250MZR/MYR and IMX253MZR/MYR address this with a global shutter function: analog memory within each pixel enables the image to be captured without the rolling-shutter distortion described in the source.

The sensors also offer controls relevant to machine-vision integration. ROI mode crops the image to selected areas, reducing the region being captured. For the IMX250MZR/MYR, up to 64 locations can be set; the source notes additional flexibility in defining and overlapping regions for this model. Trigger mode uses an external pulse to control exposure time and supports different exposure methods.

Column-parallel analog-to-digital conversion supports high-speed imaging. The source reports maximum rates, at 8-bit ADC, of up to 163.4 frames per second for the IMX250MZR/MYR and 68.3 frames per second for the IMX253MZR/MYR. Available frame rates depend on the selected operating mode and ADC setting.

Choosing a sensor configuration

The listed sensor families have a 3.45 µm by 3.45 µm unit-cell size and four-directional polarization sensing. The IMX250MZR/MYR is a 5.07-megapixel, Type 2/3 format sensor; the IMX253MZR/MYR is a 12.37-megapixel, Type 1.1 format sensor. MZR denotes monochrome versions and MYR denotes color versions; the source specifies that color capture applies to the MYR models.

When assessing a sensor for an inspection system, consider the required field of view and resolution alongside the moving-object behavior, exposure control and processing needs. Polarization outputs require subsequent signal processing, so the complete vision system—not only the sensor—must be capable of handling the intended calculations.

Practical implications for machine vision

Polarization imaging can give industrial vision systems another way to distinguish defects, materials and reflections when ordinary brightness or color images provide insufficient contrast. On-chip four-directional polarizers capture the needed measurements in one shot, while global shutter operation, ROI selection and trigger control address practical production-line requirements. The most effective applications begin with a specific inspection challenge and verify that polarization information, together with suitable processing, improves the system’s ability to identify it.

Access Official Technical PDF

About The Author