New Sony Patent Corrects Blurry Edges So Cameras Detect Fast Movement More Accurately

By Patentlyze Team · Updated Jul 24, 2026

Sony is patenting a camera chip that houses two different kinds of pixels side by side: one for capturing regular images, and one for detecting motion. The interesting twist is a built-in choice about whether to fix lens-curve errors on the motion data, a trade-off that matters a lot for speed-sensitive applications.

What Sony's two-pixel sensor setup actually does

Imagine a camera that does two jobs at once: it takes normal photos and it watches for anything that moves, almost like a motion alarm built into the sensor itself. That second job is handled by special pixels that don't record full images; they just fire a signal the instant the light hitting them changes.

The problem is that most camera lenses naturally bend the image a little at the edges, like how a fisheye photo warps straight lines. Sony's patent describes a sensor chip that can choose, on the fly, whether to mathematically straighten out that bend for the motion-detection pixels. Sometimes you want the correction for accuracy; sometimes you skip it because fixing the distortion takes extra time and the motion data is already going stale.

The chip handles both types of pixels together in one sensor array, with a built-in processing circuit that makes the call about whether distortion correction is worth it in that moment.

How the chip decides when to correct motion-pixel data

The patent covers a solid-state imaging device (the chip inside a camera) that integrates two distinct pixel types into a single sensor array:

• Standard pixels ("first pixels"): These record brightness levels to build conventional images, just like any camera sensor.
• Event-detection pixels ("second pixels"): These are inspired by so-called "event cameras," a technology borrowed from robotics and high-speed vision research. Instead of capturing full frames, each pixel independently fires a signal only when the light hitting it changes above a threshold. The result is extremely low-latency motion data.

The core of the invention is a signal processing circuit that can selectively apply distortion correction to the event-detection pixel outputs. Distortion correction (the math that straightens out the warping that wide-angle lenses introduce) is computationally non-trivial. Because event pixels generate data continuously and asynchronously, applying correction every time adds delay. The patent lets the system decide whether to run the correction or skip it depending on the use case.

Both pixel types share the same optical system, meaning they look through the same lens, which is what makes the distortion question relevant in the first place.

What this means for fast-moving cameras and robotics

Event cameras are already used in drones, industrial robots, and high-speed tracking systems because they react to motion far faster than conventional frame-based cameras. Putting event pixels and regular image pixels on the same chip, with intelligent distortion handling, could let a single sensor serve both precise imaging and ultra-fast motion detection without needing two separate optical systems.

For you as a consumer, this kind of technology could eventually show up in security cameras, automotive sensors, or even smartphone cameras that need to track fast motion without blur. The distortion-correction toggle is a small but telling detail: it shows Sony is thinking about real-world latency trade-offs, not just sensor specs on paper.

Editorial commentary on a publicly published patent application. Not legal advice.