Quote the operative limitation: a "spaced optical collimator array" sitting between the display and the optical sensor array. Goodix's grant US10331939B2, "Multi-layer optical designs of under-screen optical sensor module having spaced optical collimator array and optical sensor array for on-screen fingerprint sensing" (issued June 25, 2019; assignee Shenzhen Goodix Technology), is a granted patent, and that collimator array is what makes the whole thing function.
Here is the problem the claim solves. To read a fingerprint through a phone screen, you shine light at the finger pressed on the glass and capture the reflection with a sensor under the display. But light reflecting off the ridges and valleys of a finger scatters in every direction, and a sensor that collected all of it would see a smeared, useless blur. You need to collect only the light traveling nearly straight up — the rays that actually carry the ridge pattern. That is what a collimator does: it is an array of tiny straight channels that pass vertical light and block oblique light.
So the element that does the work is the collimator array specifically. The display, the light source, and the sensor are all conventional. The invention is interposing a structured collimator that turns scattered reflection into a clean image. Read the claim and the scope tracks that: it reads on under-display optical sensors that use a collimator-based optical path, which is a large share of the optical (as opposed to ultrasonic) under-display fingerprint market.
The granted-patent point matters for the assertion read. This is a 2019 B2 grant — issued, enforceable scope, with a priority chain that predates the broad commercialization of in-display sensors. A foundational, early, granted claim on the collimator approach is exactly the kind of asset that anchors a licensing program across the many phone makers who adopted optical in-display sensing.
Where the scope ends: ultrasonic under-display sensors — the approach Qualcomm pushed — work on a completely different physical principle (sound, not light) and do not read on a collimator-optics claim. And later refinements (anti-spoofing, moiré mitigation) live in separate filings, including Goodix's own. The discipline is to claim exactly the collimator optical path and not overstate it into "the in-display fingerprint sensor." Within the optical lane, though, this is the load-bearing grant.