Improving Fast Auto-Focus with Event Polarity

Yuhan Bao,Lei Sun,Yuqin Ma,Diyang Gu,Kaiwei Wang

DOI: https://doi.org/10.1364/OE.489717

2023-07-03

Abstract:Fast and accurate auto-focus in adverse conditions remains an arduous task. The emergence of event cameras has opened up new possibilities for addressing the challenge. This paper presents a new high-speed and accurate event-based focusing algorithm. Specifically, the symmetrical relationship between the event polarities in focusing is investigated, and the event-based focus evaluation function is proposed based on the principles of the event cameras and the imaging model in the focusing process. Comprehensive experiments on the public event-based autofocus dataset (EAD) show the robustness of the model. Furthermore, precise focus with less than one depth of focus is achieved within 0.004 seconds on our self-built high-speed focusing platform. The dataset and code will be made publicly available.

Computer Vision and Pattern Recognition,Optics

What problem does this paper attempt to address?

The paper aims to address the problem of fast and accurate autofocus under adverse conditions such as extremely dark environments, high-speed motion scenes, and high dynamic range scenes. Specifically, the paper proposes a new high-speed precise autofocus algorithm based on an event camera. By studying the symmetrical relationship of event polarity during the focusing process and based on the working principle and imaging model of the event camera, the concept of Event Polarity Rate (EPR) is proposed, and an evaluation function is constructed to assess the symmetry of these sequences. Experimental results show that the proposed Polarity-Based Focusing (PBF) method outperforms the existing Event Gradient Squared (EGS) method on the public dataset EAD, with improvements of 185 times in processing speed and 17.5% in accuracy. Additionally, the method achieves a focusing error of less than one depth of field on the self-built high-speed focusing dataset EvHF, with a processing time of less than 0.004 seconds, demonstrating its potential and advantages in practical applications.