Monitoring Camera Image Stitching Using Estimated Sampling Points

Solution Overview

Problem

Conventional monitoring camera systems face reduced stitching accuracy due to small overlapped areas between adjacent camera views, which limits the field of view and fails to meet user demands for large-range monitoring images.

Innovation Solution

An image calibrating method that combines actual and estimated sampling points to compute shifts and rotations between overlapping images, allowing for the enlargement of the overlapped area and increasing the number of sampling points for accurate image stitching.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If the overlapped area between adjacent camera views is enlarged, then the amount of feature in the overlapped area increases and stitching accuracy improves, but the field of view of the stitching image cannot contain the field of view of the predefined large-range monitoring image

Engineering Contradiction:

Improvestitching accuracyVSAvoidfield of view of stitching image

Core Design Contradiction:

Measurement precisionVSArea of stationary object

Solution Approach 1:

The system performs preliminary calibration by detecting marking points and computing traces before actual image stitching. By pre-establishing the relationship between marking points and estimating points, and calculating shift and rotation parameters in advance, the system enables accurate stitching without requiring large overlapped areas in the final image composition

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The patent introduces marking points and estimating points as intermediary elements to bridge the calibration process. These points serve as mediators between the camera views, allowing the system to compute transformation parameters (shift and rotation) without relying on extensive feature overlap in the final stitched image

Inventive Principle:

Principle #24Intermediary (Mediator)

2Area of stationary object

If the overlapped area between adjacent camera views is kept small, then the field of view of the stitching image can contain the predefined large-range monitoring image, but the amount of feature in the overlapped area is reduced and stitching accuracy decreases

Engineering Contradiction:

Improvefield of view of stitching imageVSAvoidstitching accuracy

Core Design Contradiction:

Area of stationary objectVSMeasurement precision

Solution Approach 1:

The system performs preliminary calibration by detecting marking points and computing traces before actual image stitching. By pre-establishing the relationship between marking points and estimating points, and calculating shift and rotation parameters in advance, the system enables accurate stitching without requiring large overlapped areas in the final image composition

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The patent replaces the traditional mechanical approach of relying on physical image overlap for feature detection with a computational approach. Instead of requiring extensive overlapped regions to provide enough features, the system uses marking points and trace computation to derive transformation parameters, substituting computational geometry for mechanical overlap requirements

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

3Device complexity

If only actual marking points within the overlapped area are used for computing shift, then the computation is simple, but the stitching accuracy is limited by the small number of sampling points

Engineering Contradiction:

Improvecomputation complexityVSAvoidstitching accuracy

Core Design Contradiction:

Device complexityVSMeasurement precision

Solution Approach 1:

The system performs preliminary calibration by detecting marking points and computing traces before actual image stitching. By pre-establishing the relationship between marking points and estimating points, and calculating shift and rotation parameters in advance, the system enables accurate stitching without requiring large overlapped areas in the final image composition

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The patent creates copying relationships between marking points and estimating points. For each marking point detected in one camera view, a corresponding estimating point is generated in the other view based on computed traces. This copying approach multiplies the effective number of sampling points from a limited set of actual marking points, improving accuracy without proportionally increasing computation complexity

Inventive Principle:

Principle #26Copying

Data Source

PatentUS11049281B2Image calibrating method and related monitoring camera system

Publication Date: 2021.06.29 VIVOTEK INC

AI summary

An image calibrating method is applied to a first monitoring image and a second monitoring image partly overlapped with each other. The image calibrating method includes detecting a plurality of first marking points and second marking points about a target object on the first monitoring image and the second monitoring image, computing a first trace and a second trace formed by the first marking points and the second marking points, setting a plurality of first estimating points and second estimating points on stretching sections on the first trace and the second respectively within the second monitoring image and the first monitoring image, and utilizing the first marking points and the second estimating points and/or the first estimating points and the second marking points to compute a shift between the first monitoring image and the second monitoring image.