Distributed Image Analysis via Result Set Extraction

Solution Overview

Problem

Existing video surveillance systems face challenges in image collision analysis due to incompatible AI service algorithms between provincial and prefectural servers, leading to inefficient use of computing resources, high network bandwidth requirements, and prolonged analysis times.

Innovation Solution

A distributed image analysis method and system where a first server sends image collision analysis results to a second server in the form of a result set, allowing the second server to perform further analysis and fully utilize computing power at all levels, reducing resource pressure and analysis time.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If prefectural servers upload local surveillance images to provincial central server for unified image collision analysis, then image collision analysis can be performed across different AI service algorithms, but computing and storage resource pressure on provincial center increases tremendously

Engineering Contradiction:

Improvecompatibility of image collision analysis resultsVSAvoidcomputing and storage resources

Core Design Contradiction:

Adaptability or versatilityVSQuantity of substance

Solution Approach 1:

The patent extracts only the essential analysis results (detection objects, coordinates, confidence values) from the image collision analysis process, separating them from the original surveillance images. This allows the provincial center to receive and process only the necessary data for cross-prefecture analysis without handling bulky image files, thereby reducing computing and storage resource pressure while maintaining compatibility.

Inventive Principle:

Principle #2Taking out (Extraction)

Solution Approach 2:

The patent segments the image collision analysis process into two stages: first, each prefectural server performs local analysis and uploads results; second, the provincial center performs coordinated analysis across multiple prefectures. This segmentation allows distributed computing at the prefecture level, reducing the burden on the provincial center while achieving system-wide compatibility.

Inventive Principle:

Principle #1Segmentation

2Adaptability or versatility

If provincial center receives and stores surveillance images from prefectures for image collision analysis, then unified analysis can be performed, but transmission bandwidth requirements and construction costs increase

Engineering Contradiction:

Improveunified image collision analysis capabilityVSAvoiddata transmission volume

Core Design Contradiction:

Adaptability or versatilityVSQuantity of substance

Solution Approach 1:

The patent extracts only the essential analysis results (detection objects, coordinates, confidence values) from the image collision analysis process, separating them from the original surveillance images. This allows the provincial center to receive and process only the necessary data for cross-prefecture analysis without handling bulky image files, thereby reducing computing and storage resource pressure while maintaining compatibility.

Inventive Principle:

Principle #2Taking out (Extraction)

3Measurement precision

If provincial center performs image collision analysis on large volume of surveillance images, then comprehensive analysis results can be obtained, but analysis time increases significantly

Engineering Contradiction:

Improvecomprehensiveness of analysis resultsVSAvoidimage collision analysis time

Core Design Contradiction:

Measurement precisionVSLoss of time

Solution Approach 1:

The patent applies preliminary action by having each prefectural server perform local image collision analysis before uploading results to the provincial center. This preliminary processing filters out irrelevant data and prepares results in advance, so the provincial center only needs to perform coordinated analysis across prefectures, significantly reducing total analysis time while maintaining comprehensive results.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The patent segments the image collision analysis process into two stages: first, each prefectural server performs local analysis and uploads results; second, the provincial center performs coordinated analysis across multiple prefectures. This segmentation allows distributed computing at the prefecture level, reducing the burden on the provincial center while achieving system-wide compatibility.

Inventive Principle:

Principle #1Segmentation

Data Source

PatentUS12294813B2Distributed image analysis method and storage medium for performing image analysis based on image analysis results from servers

Publication Date: 2025.05.06 HUAWEI TECH CO LTD

AI summary

A distributed image analysis method performed by a distributed image analysis system comprising a plurality of first servers and a second server. The distributed image analysis method includes: obtaining, by each of the plurality of first servers, a result set through image collision analysis, where the result set includes an index image that records a result object, the index image corresponds to an object frequency; separately sending, by each of the plurality of first servers, a result set to the second server; performing, by the second server, feature extraction on the index image in each of the result sets received from the plurality of first servers, to obtain a feature value of the index image; performing, by the second server, image collision analysis on the extracted feature value of the index image in each of the result sets, to obtain a confidence of the index image; and when determining that a confidence between the index images in the result sets received from the plurality of first servers is greater than or equal to a preset value, obtaining, by the second server, a sum of object frequencies corresponding to the index images.