Panoramic and Detail Camera Target Tracking System

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing methods for extracting target features from video data using wide-angle and telephoto cameras face challenges such as camera shake, illumination variations, and multiple targets being tracked as a whole, leading to inaccurate extraction of target attributes.

Innovation Solution

A method and apparatus that utilize a panoramic camera to capture video data, apply target detection algorithms to identify tracking targets, allocate them to detail cameras for close-up video acquisition, and use a target scoring model to select and extract attribute information from the highest scoring frames, while establishing and managing target indices to improve accuracy and reduce duplicates.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If background modeling is used to extract targets from wide-angle camera video data, then target extraction can be performed, but false targets are captured during sudden situations (camera shake, illumination variation, sudden change in capturing scene), affecting accuracy and robustness

Engineering Contradiction:
Improvetarget extraction reliabilityVSAvoidtarget extraction precision
Core Design Contradiction:
ReliabilityVSMeasurement precision

Solution Approach 1:

The system performs preliminary action by using the wide-angle camera to capture overall scene information and detect potential targets before the telephoto camera tracks them. This preliminary detection phase allows the system to establish initial target candidates that are then verified during the tracking phase, preventing false targets from being pursued. The preliminary action of broad-area scanning followed by focused verification resolves the contradiction by maintaining reliability through comprehensive coverage while improving precision through selective verification.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The system implements feedback by continuously monitoring target tracking results and using this information to adjust the wide-angle camera's detection parameters and the telephoto camera's tracking behavior. When false targets are detected during tracking (such as during camera shake or illumination changes), the feedback mechanism allows the system to re-evaluate target validity and adjust detection sensitivity, thereby maintaining both reliability and precision dynamically.

Inventive Principle:
Principle #23Feedback

2Productivity

If all moving targets in video data are extracted, then comprehensive target detection is achieved, but when multiple targets are adhered together, the telephoto camera tracks them as a whole target, preventing accurate extraction of individual target features and attributes

Engineering Contradiction:
Improvetarget detection productivityVSAvoidtarget feature extraction precision
Core Design Contradiction:
ProductivityVSMeasurement precision

Solution Approach 1:

The system applies segmentation by using the wide-angle camera to detect and separate multiple adhered targets into individual target candidates before the telephoto camera begins tracking. The wide-angle view allows the system to identify boundaries between adjacent targets and allocate them to different tracking instances. This segmentation approach enables the telephoto camera to track each target separately, maintaining both high productivity through comprehensive detection and high precision through individual feature extraction.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The system transitions from two-dimensional tracking to three-dimensional spatial reasoning by using the wide-angle camera's broader field of view to establish depth and spatial relationships between adhered targets. This dimensional change allows the system to distinguish between targets that appear merged in the telephoto camera's narrow field of view, enabling accurate separation and individual tracking of multiple targets while maintaining high detection productivity.

Inventive Principle:
Principle #17Another dimension (Dimensionality change)

3Duration of action of stationary object

If telephoto camera tracks and captures adhered targets as a whole, then tracking continuity is maintained, but features and attributes of each individual target cannot be accurately extracted

Engineering Contradiction:
Improvetracking continuityVSAvoidtarget attribute information
Core Design Contradiction:
Duration of action of stationary objectVSLoss of information

Solution Approach 1:

The system performs preliminary action by using the wide-angle camera to detect and separate adhered targets into individual candidates before the telephoto camera begins tracking. This preliminary separation ensures that each tracked target corresponds to a specific individual, preventing the loss of target attribute information while maintaining tracking continuity throughout the monitoring period.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The wide-angle camera acts as an intermediary that bridges the gap between comprehensive scene coverage and detailed target tracking. It provides the initial target separation and identification information that enables the telephoto camera to maintain continuous tracking of individual targets without losing attribute information, even when targets are initially adhered together in the scene.

Inventive Principle:
Principle #24Intermediary (Mediator)

Data Source

PatentEP3648448B1Target feature extraction method and device, and application system
Publication Date: 2024.06.19 HANGZHOU HIKVISION DIGITAL TECHNOLOGY CO LTD
  • EP3648448B1 patent drawingFigure 1
  • EP3648448B1 patent drawingFigure 2~3
  • EP3648448B1 patent drawingFigure 4~6

AI summary

The embodiment of the present application provides a method, apparatus and application system for extracting a target feature. The method includes: acquiring target area video data captured by a panoramic camera; determining, by a target detection algorithm, a tracking target having a preset feature in the target area video data; allocating the tracking target to a detail camera, such that the detail camera tracks the tracking target and acquires close-up video data of the tracking target; extracting, from the close-up video data, attribute information of the tracking target. With the embodiment of the present application, attribute information of each target can be accurately extracted, improving the accuracy of extracting the attribute information of the target.