Shuffled Image CAPTCHA for Bot Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional CAPTCHA challenges are vulnerable to attacks by software bots due to advancements in optical character recognition (OCR) technology, allowing malicious entities to easily solve distorted text challenges without significant computational resources, leading to high-volume attacks.
Innovation Solution
The development of new CAPTCHA challenges that require a significantly greater amount of computational resources to solve, involving shuffled images and decoding information, which necessitate the use of an instance of an Internet browser and/or specialized hardware, such as a graphics processing unit (GPU), to reveal the concealed challenge text.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If conventional CAPTCHA challenges using distorted text images are used, then ease of operation for human users is maintained, but security against software bot attacks deteriorates due to advancements in OCR technology
Solution Approach 1:
The challenge image is divided into multiple image portions that are spatially shuffled and separated. Each portion contains only a part of the challenge text, making it impossible for OCR technology to recognize the complete text. The image portions are distributed across different locations in the rendered webpage, requiring a browser to assemble them in the correct order to reveal the challenge text.
Solution Approach 2:
The solution moves the challenge from a simple 2D image recognition problem to a multi-dimensional problem involving spatial arrangement, coordinate mapping, and assembly logic. The challenge text is encoded across multiple dimensions (different image portions at different locations), requiring computational resources to process the spatial relationships and assemble the complete text.
2Reliability
If new CAPTCHA challenges with shuffled images are implemented, then security against software bot attacks is improved by requiring extensive computational resources, but device complexity increases
Solution Approach 1:
A software intermediary (browser extension or application) is introduced to handle the complex processing of shuffled images. This intermediary automatically downloads the challenge image, divides it into portions, determines their correct spatial arrangement using coordinate information, assembles the complete challenge text, and submits the solution. This shields the user from the complexity while maintaining security.
Solution Approach 2:
The manual mechanical process of visually assembling shuffled image pieces is replaced with an automated software system that uses coordinate mapping and image processing algorithms. The software automatically determines the correct arrangement of image portions based on their coordinate information, eliminating the need for manual manipulation while maintaining the security challenge.
3Reliability
If shuffled images with multiple image portions are used, then security barrier against high-volume attacks is enhanced, but loss of time for challenge resolution increases
Solution Approach 1:
The software intermediary performs preliminary actions by automatically downloading the challenge image, dividing it into portions, and determining the correct spatial arrangement before the user needs to solve it. The coordinate information is pre-calculated and stored, allowing for rapid assembly of the challenge text without requiring the user to manually manipulate image portions.
Solution Approach 2:
The software intermediary serves itself by automatically processing the entire challenge resolution workflow without requiring continuous user intervention. The system autonomously downloads images, processes them, assembles the challenge text, and submits solutions, reducing the time burden on users while maintaining the security barrier against bots.
Data Source
AI summary
A technique for distinguishing between a human user and a software robot. The technique includes: receiving a first communication from a device different from the at least one computer; identifying, from the first communication, a request to access a web resource; generating a shuffled image that encodes a challenge text, the generating comprising: generating an initial image encoding the challenge text; dividing the initial image into image portions; and generating the shuffled image as a shuffled arrangement of the image portions; generating decoding information for revealing the challenge text encoded in the shuffled image; transmitting the shuffled image and the decoding information to the device; receiving a second communication from the device; identifying, from the second communication, a response text; and providing the device with access to the web resource based on a comparison between the challenge text and the response text.