Image Analysis Model Adjustment With Frequency-Domain Sub-Image Sizing

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Surveillance cameras face challenges in rapidly and accurately determining the focus state of detection images due to large spatial domain information requirements, which necessitate high memory and complex computations, affecting image analysis efficiency and accuracy.

Innovation Solution

The method involves dividing detection images into sub-images, transforming them to the frequency domain, analyzing frequency responses, and adjusting the sub-image size based on prediction results to optimize image analysis efficiency and accuracy.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If spatial domain information of the detection image is analyzed to determine focus state, then the focus state can be determined, but the memory capacity requirement and computation complexity increase significantly

Engineering Contradiction:

Improvefocus state determination accuracyVSAvoidmemory capacity and computation complexity

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The detection image is divided into multiple sub-images, and each sub-image is further divided into blocks. This segmentation reduces the amount of spatial domain information that needs to be processed simultaneously, thereby reducing memory capacity requirements and computation complexity while maintaining focus state determination capability

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent replaces the conventional spatial domain analysis method with a frequency domain analysis method using Fourier transform. This substitution changes the mathematical domain of processing from spatial to frequency, which reduces the computational burden and memory requirements for determining focus state

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

2Measurement precision

If conventional spatial domain analysis is used to determine focus state, then the focus state can be determined, but the computation time becomes lengthy

Engineering Contradiction:

Improvefocus state determination accuracyVSAvoidcomputation time

Core Design Contradiction:

Measurement precisionVSLoss of time

Solution Approach 1:

The patent substitutes spatial domain analysis with frequency domain analysis through Fourier transform. The frequency domain approach allows for more efficient computation of focus metrics by exploiting the periodicity and frequency characteristics of the image data, significantly reducing computation time compared to conventional spatial domain methods

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The patent changes the domain parameter from spatial coordinates to frequency coordinates. By transforming the image data into the frequency domain, the computation of focus state determination becomes more efficient as frequency-based operations require fewer computational steps than spatial domain operations for the same accuracy level

Inventive Principle:

Principle #35Parameter changes

Applied Scientific Principles

This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.

Function Achieved in This Case

This approach enables rapid and accurate determination of image focus state by optimizing sub-image size and frequency domain analysis, enhancing image analysis efficiency and accuracy.

Implementation Method 1

transforming the plurality of first sub-images from a spatial domain to a frequency domain for generating a plurality of first frequency domain images

Methodology Applied

Scientific EffectFourier transform:

Data Source

PatentUS12462522B2Image analysis model adjustment method and image analysis apparatus

Publication Date: 2025.11.04 VIVOTEK INC

AI summary

An image analysis model adjustment method is applied to an image analysis apparatus. The image analysis model adjustment method includes dividing a detection image acquired by the imager analysis apparatus into a plurality of first sub-images via a valid size, transforming the plurality of first sub-images from a spatial domain to a frequency domain to generate a plurality of first frequency domain images, distributing the plurality of first frequency domain images into several crop groups via a predefined set value, analyzing frequency responses at the same frequency in each crop group to provide a specific frequency response, collecting specific frequency responses of the crop groups to generate frequency domain group data, and analyzing the frequency domain group data to acquire a prediction result of the detection image so as to decide whether to adjust the valid size in accordance with the predicted result.