Rotating Image Segments for AI Verification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional verification code interception technologies face limitations in defense effect, are product-dependent, and negatively impact user experience, as they struggle to effectively differentiate between human and computer inputs due to advancements in AI and machine learning, particularly in image recognition and semantic understanding.
Innovation Solution
An image verification method and apparatus that utilizes two rotatable image portions with a potential gap between them, increasing the difficulty of semantic understanding and visual recognition for computers while maintaining ease for human users, by requiring alignment of these images through rotation, thereby enhancing the interception of computer attacks.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional verification code technologies are used, then the implementation is simple and cost-effective, but the interception capability against AI attacks is insufficient
Solution Approach 1:
The verification code is divided into multiple image portions (first image portion and second image portion) that can be independently rotated. This segmentation increases the complexity for AI recognition while maintaining a relatively simple implementation structure on the server side.
Solution Approach 2:
The image portions are made dynamically rotatable, transforming from static verification codes to dynamic ones. The first image portion can be rotated independently by the user, creating multiple possible configurations that are difficult for AI to predict and crack, while the implementation remains straightforward through standard image rotation operations.
2Object-affected harmful factors
If AI technology develops, then the ability to crack verification codes improves, but the security of verification codes deteriorates
Solution Approach 1:
The verification code uses asymmetric configuration where the first image portion and second image portion have different rotational characteristics. The first image portion is rotatable while the second remains fixed or has different rotation properties, creating an asymmetric puzzle that is easy for humans to solve intuitively but difficult for AI to crack due to the lack of clear recognition patterns.
Solution Approach 2:
The verification code changes multiple parameters simultaneously including rotation angles, image positioning, and gap distances. These parameter changes create a high-dimensional solution space that is difficult for AI to search efficiently, while human users can still solve it through spatial reasoning and intuitive manipulation.
3Reliability
If verification code interception technologies are enhanced, then security improves, but user experience deteriorates
Solution Approach 1:
The verification code requires only partial rotation of the first image portion to a specific angle rather than complex multi-step operations. This partial action approach maintains ease of operation for human users while the precise angular requirement and relationship with the second image portion create sufficient difficulty for AI attacks.
Data Source
AI summary
According to an embodiment of the present application, there is provided an image verification method and apparatus, an electronic device and a computer-readable storage medium, which can be applied to the technical field of network security and the field of image processing and recognition. The image verification method includes: displaying a first image portion and a second image portion, the first image portion being rotatable with respect to the second image portion, the first image portion and the second image portion being obtained by cropping an original image and rotating the cropped first image portion with respect to the second image portion; receiving an operation for rotating the first image portion; rotating the first image portion based on the operation; and determining whether an angle of the first image portion relative to the second image portion matches with the original image, in response to determining that the operation is ended.