A TVS event clustering method based on optical flow analysis

Abstract

The invention discloses a method for dividing TVS events into clusters based on optical flow analysis. The method divides the TVS events into clusters by computing the optical flows of the TVS events, extracts the accurate position and the shape information of a moving target according to the information of the clusters, and is characterized by acquiring the optical flow speed of each spot by analyzing the optical flows event by event, further dividing the events into clusters according to the similarity of speed, positions, and output time, and dynamically updating pre-existing events in the clusters according to cluster speed. The method divides the TVS events into clusters according to the generated time, the spatial position, and the optical flow speed of the events, and has advantages that the combination between a characteristic that spatial moving targets have same projection speed on a 2D imaging plane and space-time similarity may solves overlapping and shielding problems of the targets so as to achieve higher accuracy; and cluster speed may update the pre-existing events in the clusters.

Description

technical field [0001] The invention relates to multiple technical fields such as computer vision, image processing, and image sensor design, and in particular to a TVS event clustering method based on optical flow analysis. Background technique [0002] Semiconductor silicon-based image sensors (CCD and CMOS) have become the most important visible light imaging devices. Like the original silver iodide film, they all use the "frame sampling" method to complete light intensity measurement: after all pixels are synchronously reset, they begin to collect photoelectric charges. After the set exposure time is reached, the photocharge collected by each pixel is read out and converted into a voltage; the voltage becomes a digital quantity after analog-to-digital conversion, and is output and stored. The two-dimensional matrix composed of all pixel brightness values ​​is the captured image. The usual video shooting speed is 30 frames per second, that is, the charge collection time ...

AI Technical Summary

Problems solved by technology

However, this sampling method also has prominent shortcomings: first, the static background is repeatedly sampled, and the data redundancy is high, which brings great pressure to real-time image processing and transmission storage; second, the time resolution is low, and the pixels cannot distinguish the charge collection. Any change in the light intensity within a certain period of time, but only the cumulative sum of photocharges during this period is measured, so it is not conducive to the tracking and identification of high-speed moving targets

[0012] 1. Events belonging to the same target may have large discreteness in output time. The reason is that TVS uses serial AER output, and multiple events generated at the same time need to be arbitrated, and the output of events generated at two adjacent points in space at the same time The time difference is large;

[0013] 2. When there are multiple moving targets in the scene, there will be overlapping occlusion problems, and events with temporal and spatial similarities may not belong to the same target

The above methods all judge the clustering of events based on the geometric distance from the center of the cluster, so they are only suitable for objects with symmetrical and regular shapes.

Images