A spatio-temporal normalization method for event-oriented camera data processing

Abstract

The present invention provides a spatio-temporal normalization method for event-oriented camera data processing, mainly based on the property that the later the event is activated, the greater the value of its time stamp, the pixels are connected by using a doubly linked list, and the relative Timing relationship; according to the known nature of the pixel coordinate position of the activated event, use sequence update to rearrange the node sequence; in the sequence update stage, according to the needs of different applications, use synchronous sequence update and asynchronous sequence update; when selecting a sequence , according to the sequence-identical property of pixels with the same timestamp, to remove wrong sequences. The invention utilizes the unique properties of the event camera to perform SAE spatio-temporal normalization processing, and has the characteristics of self-adaptation and high efficiency.

Description

technical field [0001] The invention belongs to the field of signal processing, and in particular relates to normalizing processing of event camera data. Background technique [0002] The event camera is a dynamic visual sensor that mimics the optic nerve of the human eye and is driven by events. Different from traditional picture-based cameras, event cameras have no fixed observation frequency. Each pixel circuit independently detects the light intensity level. When the light intensity changes, it outputs an event data with a very fast response speed, making the event camera efficient It has great application potential in systems with low power consumption. However, since the output of the event camera is the timestamp information in the event data, it is difficult to directly apply traditional vision processing algorithms, so an effective spatiotemporal representation of asynchronous event streams is required. A commonly used expression is SAE (Surface of Active Events),...

AI Technical Summary

Problems solved by technology

Among the existing normalization methods, the method based on time window is easy to lose potentially useful information, and needs to manually adjust parameters according to the dynamic range, which is less robust; while the normalization method based on sorting is at the cost of running time, Inefficient when dealing with large size SAE

In the current methods of preprocessing event camera data, there is no efficient adaptive normalization method

Images