Image Matching Device Using Phase Only Correlation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current biometric authentication methods, particularly one-to-N authentication, face inefficiencies due to the need for multiple image processing devices, which increase costs and authentication time, especially when images are rotated or have different angles, leading to longer processing times and potential interference between similarity peaks.
Innovation Solution
An image matching device and method that generate a mixed image by multiplying and adding recorded images with different phase components, then calculate a complex similarity image and obtain similarity from its projected component, allowing for efficient similarity calculation without the need for multiple processing devices by using Phase Only Correlation (POC) and Fourier transforms.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If multiple image processing devices are used in parallel to reduce authentication time, then authentication speed is improved, but device cost and system complexity increase
Solution Approach 1:
The patent merges multiple image processing functions into a single device by using mathematical operations to combine multiple recorded images into a mixed image. The similarity calculation unit computes similarity between the input image and all recorded images simultaneously through complex number operations, eliminating the need for multiple parallel processing devices while maintaining high authentication speed.
Solution Approach 2:
The single image processing device is designed to perform multiple functions: it can calculate similarity with multiple recorded images, handle various rotation angles, and process different authentication scenarios (one-to-one and one-to-N authentication) all within one unified system, making the device universal and multi-functional.
2Measurement precision
If multiple recorded images are processed to improve image matching accuracy, then matching accuracy is improved, but processing time increases
Solution Approach 1:
The patent transforms the image processing problem into the complex number domain, where images are represented as complex numbers. This parameter change enables the use of efficient mathematical operations (multiplication and addition of complex numbers) that can process multiple images simultaneously, improving both accuracy and speed compared to traditional real-number-based image processing methods.
Solution Approach 2:
The patent replaces traditional mechanical or sequential image processing methods with mathematical field operations. By using complex number arithmetic and Fourier transform-based operations, the system substitutes sequential processing with parallel mathematical computations, significantly reducing processing time while maintaining high matching accuracy.
3Adaptability or versatility
If images with different rotation angles are matched to improve adaptability, then system versatility is improved, but authentication time increases
Solution Approach 1:
The patent handles rotation angle variations by transforming images into the frequency domain using Fourier transform, where rotation operations become phase shifts in the complex plane. This dimensional transformation allows the system to efficiently handle multiple rotation angles through simple phase adjustments rather than complex spatial operations, maintaining versatility while reducing processing time.
Data Source
AI summary
An image matching device includes: a mixed image generation portion generating a mixed image in an operation satisfying linearity, the mixed image being obtained by multiplying each of two or more recorded images and each of phase components of a complex plane different from each other and adding multiplied recorded images; a complex similarity image generation portion generating a complex similarity image through a similarity operation between one or more input image and the mixed image; and a similarity obtain portion obtaining similarity from a projected component of the complex similarity image toward a vector of the phase component.