Qualcomm Patents a Way to Stop Wipers from Confusing Your Car's Cameras

By Patentlyze Team · Updated May 16, 2026

Every time a windshield wiper sweeps across a dashcam's field of view, the image briefly flickers — and that flicker can confuse the AI systems keeping your car in its lane. Qualcomm has filed a patent for a smarter camera pipeline that detects wiper blades in real time and adjusts image processing accordingly.

What Qualcomm's wiper-aware camera system actually does

Imagine you're driving in a rainstorm and your car's camera-based safety system — the one that watches for pedestrians or reads lane markings — suddenly gets confused because the windshield wiper keeps interrupting its view. That's a real problem for modern driver-assistance systems, and it's surprisingly tricky to solve.

Qualcomm's patent describes a system that watches incoming camera frames and assigns a probability score to each one: how likely is it that a wiper blade is currently crossing the image? That score then controls a time filter — essentially a smoothing algorithm — that adjusts how the camera maps brightness and contrast before passing the image downstream to the vehicle's AI.

The result is that wiper-obscured frames get handled differently than clean frames, so the pipeline doesn't overreact to the sudden light change a wiper blade causes. It's a focused, practical fix for a specific annoyance in automotive imaging.

How the probability score drives the tone-mapping filter

The system works in two conceptual stages: wiper detection and tone mapping control.

In the first stage, the camera captures an image frame and a machine learning model evaluates it to produce a probability value — a number expressing how confident the system is that a windshield wiper blade appears in that frame. This isn't a binary yes/no; it's a continuous score, which means the system can respond proportionally rather than switching abruptly.

In the second stage, that probability value feeds into a time filter (a filter that considers the history of recent frames, not just the current one) which then governs tone mapping — the process of converting raw high-dynamic-range sensor data into a well-exposed output image suitable for display or downstream AI processing. When wiper probability is high, the filter is tuned to be more conservative, avoiding the sharp brightness adjustments that a wiper sweep would otherwise trigger.

The output image — the one actually used by the vehicle assistance system — comes from tone-mapping a second frame (potentially the next or a reference frame) using the filter derived from the first. This separation gives the pipeline a moment of temporal context to make smarter decisions:

• Detect wiper presence in frame N
• Compute an appropriate time filter
• Apply that filter when processing frame N+1 (or an adjacent frame)

What this means for safer ADAS cameras in bad weather

Driver-assistance cameras operate in some of the worst imaging conditions imaginable — glare, rain, sudden darkness in tunnels — and tone mapping is one of the main tools that keeps images usable across that range. The problem is that aggressive tone mapping can be destabilized by a repeating, high-contrast object like a wiper blade sweeping the frame several times per second. If the camera's brightness algorithm chases every wiper pass, the video feed becomes unstable and downstream object-detection models suffer.

Qualcomm supplies the camera ISP (image signal processor) chips that sit at the heart of many automotive camera systems, so this kind of low-level imaging fix is squarely in their product lane. A more stable image pipeline in rain means your car's lane-keeping or emergency braking system gets cleaner data when road conditions are already dangerous — which is exactly when you need it most.

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