Client-Side Image Clarity Filtering for Accurate Recognition
Find Innovative SolutionsGenerate 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
Engineering 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
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.
2Measurement precision
If coordinate system conversion is performed before defect correction, then processing accuracy is improved, but processing time increases
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.
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.
3Adaptability or versatility
If multiple coordinate systems are handled simultaneously, then versatility is improved, but device complexity increases
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.
Data Source
Figure 1
Figure 2~3
Figure 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.