Event-Based Vision Sensor Clustering for Shape-Adaptive Tracking

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing clustering methods for event-based vision sensors are inefficient in handling objects of varying sizes and shapes with high computational complexity, limiting their effectiveness in object detection and tracking applications.

Innovation Solution

A clustering method that utilizes a matrix to store indicators for each pixel, determining if it is assigned to a cluster or a candidate, and applies simple rules based on neighborhood pixels to efficiently form clusters, updating this matrix continuously to track objects without rectangular or circular limitations.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If traditional clustering methods are used to handle objects of varying sizes and shapes, then clustering accuracy is improved, but computational complexity increases significantly

Engineering Contradiction:
Improveclustering accuracyVSAvoidcomputational complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent segments the image space into discrete pixel units and processes clustering decisions locally at each pixel level. Each pixel maintains an independent indicator state, allowing parallel processing and avoiding global computational complexity while maintaining clustering accuracy through local neighborhood interactions.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

Each pixel automatically determines its own cluster assignment status by evaluating its indicator value and neighborhood indicators without requiring centralized control. The system enables self-organizing cluster formation where pixels autonomously update their states based on local rules, reducing overall computational burden.

Inventive Principle:
Principle #25Self-service

2Device complexity

If rectangular or circular tracking boxes are used, then device complexity is reduced, but adaptability to various object shapes deteriorates

Engineering Contradiction:
Improvedevice complexityVSAvoidadaptability to object shapes
Core Design Contradiction:
Device complexityVSAdaptability or versatility

Solution Approach 1:

The patent applies different cluster assignment states (unassigned, candidate, assigned) to different pixels based on their local characteristics and neighborhood contexts. This allows the clustering boundary to naturally adapt to object shapes by assigning pixels locally rather than imposing global geometric constraints, achieving shape adaptability without complex tracking structures.

Inventive Principle:
Principle #3Local quality

Solution Approach 2:

Instead of fitting objects into predefined rectangular or circular templates, the patent inverts the approach by allowing pixels to self-organize into clusters that naturally conform to object boundaries. The clustering shape emerges from pixel-level decisions rather than being constrained by external geometric forms.

Inventive Principle:
Principle #13The other way round (Inversion)

3Measurement precision

If continuous data buffering is implemented, then measurement precision is improved, but memory consumption increases

Engineering Contradiction:
Improvemeasurement precisionVSAvoidmemory consumption
Core Design Contradiction:
Measurement precisionVSQuantity of substance

Solution Approach 1:

The patent extracts only the essential clustering state information (indicator values) from the event stream and maintains minimal buffering. By extracting and maintaining only the necessary pixel state indicators rather than buffering complete event sequences, the system achieves measurement precision through state tracking while minimizing memory consumption.

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The system discards event data that has been processed and uses the recovered pixel indicator states for continuous clustering updates. Once a pixel's cluster status is determined, the raw event data can be discarded while the essential clustering information is retained in the indicator, reducing memory requirements while maintaining measurement precision.

Inventive Principle:
Principle #34Discarding and recovering

Data Source

PatentEP4033398B1A clustering method applied to a change event stream produced by an event-based vision sensor
Publication Date: 2026.02.11 TOYOTA JIDOSHA KK
  • EP4033398B1 patent drawingFigure 1
  • EP4033398B1 patent drawingFigure 2A~2D
  • EP4033398B1 patent drawingFigure 3

AI summary

A clustering method applied to a change event stream produced by an event-based vision sensor which comprises a plurality of pixels and is configured to output a stream of change events comprising: providing (S00) a matrix having the same dimensions as the event-based vision sensor in which at each position, an indicator is stored to indicate if the corresponding pixel of the event-based sensor was not used or assigned to a given cluster or a candidate to form a cluster, and when a change event is detected (S01) in the change event stream at a pixel: determining (S04) a group of pixels centered on the pixel, modifying (S07) the indicator of the pixel if a predetermined rule based on at least the indicators of the pixels of the group of pixels is verified. The invention also concerns a corresponding system.