Vehicle Counting via Compressed Video Motion Vectors

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing vehicle counting methods, such as roadway sensors and traditional video-based systems, are either inaccurate, expensive, or computationally expensive, and lack efficiency in real-time data processing, especially in bandwidth-constrained environments.

Innovation Solution

A method and system that utilize motion vectors from compressed video streams to count vehicles directly on the compressed data stream, allowing for real-time vehicle counting without decompression, and optionally embedding count results in metadata within the compressed stream.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If traditional video-based vehicle counting is performed offline on uncompressed video streams, then measurement precision is improved, but productivity deteriorates due to high computational cost and time consumption

Engineering Contradiction:
Improvevehicle counting accuracyVSAvoidprocessing speed
Core Design Contradiction:
Measurement precisionVSProductivity

Solution Approach 1:

The patent applies preliminary action by performing vehicle counting analysis during the video compression process itself, rather than as a separate post-processing step. The motion vector extraction and vehicle detection algorithms are integrated into the compression pipeline, so that counting information is generated while the video is being compressed and stored, eliminating the need for separate offline processing and significantly improving processing efficiency

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent extracts motion vectors from the compressed video stream without requiring full decompression. By utilizing the motion vectors already generated during compression and selectively processing only relevant portions of the compressed data, the system achieves accurate vehicle counting while maintaining high processing speed and low computational overhead

Inventive Principle:
Principle #2Taking out (Extraction)

2Productivity

If roadway sensors are used for vehicle counting, then productivity is improved through real-time data collection, but device complexity and cost increase

Engineering Contradiction:
Improvereal-time data collection capabilityVSAvoidsystem complexity
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The patent makes existing video surveillance cameras multi-functional by enabling them to perform both their original surveillance function and vehicle counting function simultaneously. By analyzing motion vectors from the video streams already being captured for surveillance purposes, the system eliminates the need for separate dedicated vehicle counting devices like roadway sensors, thereby reducing device complexity and deployment cost while maintaining real-time counting capability

Inventive Principle:
Principle #6Universality (Multi-functionality)

Solution Approach 2:

The patent enables video compression systems to serve themselves by generating vehicle counting information as a byproduct of the compression process. The motion vectors required for compression are reused for vehicle detection and counting, so the same computational work serves dual purposes without requiring additional processing resources or specialized hardware

Inventive Principle:
Principle #25Self-service

Data Source

PatentUS9424747B2Vehicle counting methods and systems utilizing compressed video streams
Publication Date: 2016.08.23 CONDUENT BUSINESS SERVICES LLC
  • US9424747B2 patent drawing
  • US9424747B2 patent drawing
  • US9424747B2 patent drawing

AI summary

This disclosure provides method and systems of recording a predetermined event associated with a moving object, the predetermined event captured with an image capturing unit and one or more of the associated frames compressed, producing one or more motion vectors. According to one exemplary embodiment, vehicle counting is performed based on motion vectors produced during the data compression process, either inline or offline.