Client-Side Image Clarity Filtering for Accurate Recognition

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing image identification systems face challenges in improving accuracy and efficiency due to the inclusion of images that do not meet clarity conditions, leading to invalid inputs and increased network resource usage.

Innovation Solution

An image processing method that determines a target image with preset clarity conditions before identification, using client-side models to filter images and reduce invalid inputs, and performs initial identification on the client to enhance accuracy and efficiency.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Speed

If image data is corrected in the original coordinate system without conversion, then processing speed is improved, but processing accuracy deteriorates due to oblique projection distortion

Engineering Contradiction:
Improveimage processing speedVSAvoidimage processing accuracy
Core Design Contradiction:
SpeedVSMeasurement precision

Solution Approach 1:

The image processing is divided into two distinct stages: first, coordinate system conversion is performed on the input image data to transform it from an oblique projection coordinate system to a orthogonal projection coordinate system; second, defect correction processing is performed on the converted image data. This segmentation allows each stage to be optimized independently, resolving the contradiction between speed and accuracy.

Inventive Principle:
Principle #1Segmentation

2Measurement precision

If coordinate system conversion is performed before defect correction, then processing accuracy is improved, but processing time increases

Engineering Contradiction:
Improvedefect detection accuracyVSAvoidtotal processing time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

Coordinate system conversion is performed as a preliminary action before defect correction processing. By pre-converting the image data to the orthogonal projection coordinate system, the subsequent defect correction can be performed with high accuracy without repeatedly handling coordinate transformations during the correction process itself.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent replaces complex real-time coordinate transformation mechanics during defect correction with a simpler approach: perform the transformation once beforehand, then work with the pre-converted data. This substitution of the processing sequence reduces computational overhead and total processing time while maintaining accuracy.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

3Adaptability or versatility

If multiple coordinate systems are handled simultaneously, then versatility is improved, but device complexity increases

Engineering Contradiction:
Improvecoordinate system compatibilityVSAvoidprocessing system complexity
Core Design Contradiction:
Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The coordinate system conversion function is extracted as a separate, independent preprocessing stage from the defect correction processing. This extraction allows the main defect correction algorithm to work with a single, standardized coordinate system (orthogonal projection), simplifying its logic while still supporting multiple input coordinate types through the separate conversion module.

Inventive Principle:
Principle #2Taking out (Extraction)

Data Source

PatentEP4325439B1Image processing method and apparatus, and device and storage medium
Publication Date: 2026.05.20 BEIJING ZITIAO NETWORK TECH CO LTD
  • EP4325439B1 patent drawingFigure 1
  • EP4325439B1 patent drawingFigure 2~3
  • EP4325439B1 patent drawingFigure 4

AI summary

Provided in the present disclosure are an image processing method and apparatus, and a device and a storage medium. The method comprises: when a preset recognition operation triggered for a photographing picture on a photographing page is received, determining, on the basis of the photographing picture on the photographing page, a target image which has an image definition that meets a preset definition condition; and then sending the target image to an image recognition serving end to perform recognition on the target image. It can be seen that, by means of the image processing method provided in the embodiments of the present disclosure, a target image, which has an image definition that meets a preset definition condition, can be determined first before image recognition is performed, such that an image recognition serving end can perform recognition on the basis of the target image which has the image definition that meets the condition, thereby improving the accuracy of image recognition.