IMU-enhanced event camera optical flow estimation method and system

Abstract

The invention provides an IMU-enhanced event camera optical flow estimation method and a system, and the method comprises the steps: firstly reconstructing a fuzzy brightness image at any moment intoa clear brightness image through an EDI model, building a basic optical flow model through combining with a brightness constant hypothesis, and enabling the EDI to represent event-based double integration; then adding the IMU as a constraint into a basic optical flow model to realize continuous optical flow estimation in any motion consistent scene; separating a background and a foreground for processing when a hybrid motion scene appears, introducing IMU constraints during background optical flow estimation, introducing sparse constraints during foreground optical flow estimation, jointly estimating the background optical flow and the foreground optical flow of the hybrid motion scene through an alternate iterative updating mode, and finally, the overall continuous optical flow of the scene is obtained through combination. The method is suitable for scenes comprise a motion consistency scene in which a single background moves or a foreground object moves consistently, and a mixed motion scene in which the background and the foreground move in different directions and sizes.

Description

technical field [0001] The invention belongs to the field of image processing, and in particular relates to a technical solution for continuous optical flow estimation in high dynamic range and high-speed motion scenes. Background technique [0002] In the field of computer vision, optical flow estimation has always been a core technical issue and plays an important role in applications such as navigation, motion segmentation, tracking, and image registration. The event camera is a new type of bionic sensor, as attached figure 1 As shown, the camera has independent pixels that asynchronously generate a series of pulses called "events" only when light intensity changes occur. . Since the event camera samples events at a dynamically changing rate of the scene, it has several advantages over traditional optical cameras: high time resolution, low latency, high dynamic range (High Dynamic Range, HDR), low power consumption and low bandwidth. Based on the advantages above, the...

AI Technical Summary

Problems solved by technology

[0004] But usually, a small number of IMUs carry too little information, and no suitable technical solutions have appeared yet

Images