Monday, December 18, 2023

OmniVision 15MP/1MP hybrid RGB/event vision sensor (ISSCC 2023)

Guo et al. from Omnivision presented a hybrid RGB/event vision sensor in a paper titled "A 3-Wafer-Stacked Hybrid 15MPixel CIS + 1 MPixel EVS with 4.6GEvent/s Readout, In-Pixel TDC and On-Chip ISP and ESP Function" at ISSCC 2023.

Abstract: Event Vision Sensors (EVS) determine, at pixel level, whether a temporal contrast change beyond a predefined threshold is detected [1–6]. Compared to CMOS image sensors (CIS), this new modality inherently provides data-compression functionality and hence, enables high-speed, low-latency data capture while operating at low power. Numerous applications such as object tracking, 3D detection, or slow-motion are being researched based on EVS [1]. Temporal contrast detection is a relative measurement and is encoded by so-called “events” being further characterized through x/y pixel location, event time-stamp (t) and the polarity (p), indicating whether an increase or decrease in illuminance has been detected.


Schematic of dual wafer 4x4 macro-pixel and peripheral readout circuitry on third wafer.

EVS readout block diagram and asynchronous scanner with hierarchical skip-logic.
Event-signal processor (ESP) block diagram and MIPI interface.

Sensor output illustrating hybrid CIS and EVS data capture. 10kfps slow-motion images of an exploding water balloon from 1080p, 120fps + event data.
Characterization results: Contrast response, nominal contrast, latency and noise vs. illuminance.

Technology trend and chip micrograph.


  1. Thanks for posting. Finally these industrial event cameras have come back full circle, frames, then events, now frames+events. This chip and the one from Sony both seem to be highly integrated designs from skillful industrial teams.

  2. I'm curious to know how they merge a rolling shutter RGB image with a globally detected events, especially with scenes containing fast motions which are almost always the case with slow motion capture.


All comments are moderated to avoid spam and personal attacks.