ÃÈÃÃÉçÇø

Researchers have developed a real-time imaging system that can capture images of fast-spinning objects over long durations. Real-time monitoring of rotating parts such as the turbine blades used in power plants or the fan blades of jet engines is critical for detecting early signs of damage—such as wear or cracks—helping prevent serious failures and reducing maintenance needs.

"Capturing clear images of fast-spinning objects is challenging because they tend to blur or look grainy," said research team member Zibang Zhang from Jinan University in China. "Although high-speed cameras can help, they're expensive and can't be used for long periods. Our method overcomes this challenge by virtually freezing time by exploiting the repetitiveness of the object's motion."

In the journal Optics Letters, the researchers their new imaging system, which is based on a single-pixel detector. They show that it can capture images of an object spinning at around 14,700 rounds per minute (rpm).

"The system could detect wear or cracks that can develop in high-speed metal cutting and grinding tools over time—without stopping the machines—improving safety and extending the equipment's lifespan," said Zhang. "In the future, this technology could be integrated into smart manufacturing systems, aircraft maintenance platforms or even home appliances like car engines, blenders, fans, air conditioners and hard drives, making these devices smarter and safer."

Freezing time

Imaging fast-spinning objects is not an easy task for conventional imaging techniques because rotation leads to severe motion blur. Although reducing exposure time can ease the problem, this tends to create noisy images because fewer photons are captured. High-speed cameras can also be used to capture fast-spinning objects but are prohibitively expensive and cannot be used continuously over long time periods.

As part of a project aimed at developing an optical system for online engine inspection, the researchers developed a new system that overcomes many of the challenges of imaging fast-spinning objects by using structured illumination and single-pixel detection.

This imaging approach projects patterned light onto a scene and captures the resulting intensity variations with a single-pixel detector, allowing a computer to reconstruct a detailed image without needing a traditional camera sensor. Single-pixel detectors are photodiodes with only one pixel available. Compared to traditional camera sensors, single-pixel detectors have higher sensitivity, a wider dynamic range and faster response, making them suitable for imaging fast-spinning objects.

"The key to the method is synchronization, which essentially freezes time by keeping the target object stationary compared to the pattern projection," explained Zhang. "By using synchronized illumination, we converted a dynamic imaging problem into a static imaging problem."

The imaging system captures clear images of rotating objects by aligning with their repetitive motion. This is much like painting a sunrise over several days, where each time the sun rises, the artist paints a small part of the scene. Even though the sun moves continuously, it is possible to capture the whole picture by setting an alarm to sync with the sun's daily return.

Capturing a spinning object

For the imaging setup, the researchers used a high-speed projector—specifically, a digital micromirror device enabling projection speeds of up to 22,000 Hz—to illuminate the rotating object with a series of patterns. The single-pixel detector acquires a measurement for each pattern projection. Once the object spins around once, the projector switches to the next pattern.

To synchronize the projection, the researchers "set an alarm clock" by aiming a laser at one blade of the spinning object, creating backscattered pulses. When the number of pulses matches the number of blades, it signals the projector to switch patterns—like setting off the alarm—allowing clear imaging of the rotating object using just a single-pixel detector.

To demonstrate the system, the researchers showed that it could reconstruct real-time high-quality still images of a model jet engine 11 cm in diameter rotating at about 2,170 rpm and a CPU cooling fan rotating at approximately 14,700 rounds per minute. The researchers said that the method, which requires no prior knowledge of an object, can also be used to image objects with unstable rotation speed.

Next, the team plans to improve the system's portability and make it easier to integrate into a real aircraft engine.