Method and system for removing background noise and flicker noise in flicker scene step by step

By combining light intensity information and camera angular velocity information, using neural networks and filter technology, the problem that event cameras are difficult to remove background noise and flicker noise in the flicker light source environment is solved, significantly improving image quality.

CN120201325APending Publication Date: 2025-06-24BEIJING WUZI UNIVERSITY
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202510366807.5
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2025-03-26
Publication Date
2025-06-24

AI Technical Summary

Technical Problem

In the flickering light source environment, it is difficult for event cameras to effectively remove background noise and flicker noise, resulting in a degradation of image quality.

Method used

By combining the intensity information of light and the camera angular velocity information acquired by the IMU, the possibility of an event occurring at each pixel at each moment is marked. Then, the event data is preprocessed using the trained background noise removal neural network, estimate the true brightness change threshold and correct the event data. Finally, the flickering noise in the corrected event data is removed using an improved feedforward comb filter.

Benefits of technology

Effectively removes background noise and flicker noise, improving the signal-to-noise ratio of event data, and is especially suitable for fluorescent and LED lighting environments.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN120201325A_ABST
    Figure CN120201325A_ABST
Patent Text Reader

Abstract

The invention provides a method and system for removing background noise and flicker noise in a flicker scene step by step, and the method comprises the steps: marking the possibility that each pixel has an event at each moment through the intensity information of light in the scene and the angular velocity information of a camera obtained by an IMU (Inertial Measurement Unit); acquiring event data generated by an event camera, wherein the event data comprises a spatial position, a timestamp and polarity of an event; taking the difference between the timestamp of the event recently generated by each pixel and the timestamp of the marked event as an input feature vector, encoding the input feature vector into a plurality of channels, and inputting the trained background noise removal neural network to remove background noise; estimating a real brightness change threshold value cp and a deviation bp, and correcting the event from which the background noise is removed through the estimated cp and bp, the corrected event comprising flicker noise; modeling the corrected event into a time domain signal; and using a flicker noise removal filter to remove flicker noise in the corrected event. According to the invention, background noise and flicker noise can be removed step by step.
Need to check novelty before this filing date? Find Prior Art