Search

KR-20260067147-A - ELECTRONIC DEVICE FOR IMPROVING SINGULARITY THROUGH FILTER-IN OF NOISY IMAGE PORTION AND METHOD OF OPERATING THE SAME

KR20260067147AKR 20260067147 AKR20260067147 AKR 20260067147AKR-20260067147-A

Abstract

According to various embodiments, an electronic device for singularity enhancement through a noise image portion filter-in includes a processor, wherein the processor is configured to i) divide first image data containing a fingerprint into a plurality of preset blocks, ii) obtain a Fast Fourier Transform (FFT) spectrum by applying a Discrete Time Fourier Transform (DFT) to one of the divided plurality of preset blocks, iii) output reconstructed image data by applying an inverse FFT after applying a designed filter to the FFT spectrum, and execute operations i), ii), and iii) on all of the preset plurality of blocks. According to various embodiments, a method of operation of an electronic device for singularity enhancement through a noise image portion filter-in comprises: i) dividing first image data containing a fingerprint into a plurality of predetermined blocks; ii) obtaining a Fast Fourier Transform (FFT) spectrum by applying a Discrete Time Fourier Transform (DFT) to one of the divided plurality of predetermined blocks; and iii) outputting reconstructed image data by applying an inverse FFT after applying a designed filter to the FFT spectrum; and may further comprise, after the step of outputting the reconstructed image data, repeating steps i), ii), and iii) for all of the plurality of predetermined blocks. Various other embodiments are also possible.

Inventors

  • 이상현

Assignees

  • 호남대학교 산학협력단

Dates

Publication Date
20260512
Application Date
20241105

Claims (6)

  1. In an electronic device for singularity enhancement through a noise image portion filter-in, Includes a processor, The above processor is, i) Divide the first image data containing the fingerprint into a plurality of pre-set blocks, and ii) Obtain a Fast Fourier Transform (FFT) spectrum by applying Discrete Time Fourier Transform (DFT) to one of the aforementioned pre-set divided blocks, and iii) After applying the designed filter to the above FFT spectrum, apply the inverse FFT to output the reconstructed image data, and Set to execute the operations i), ii), and iii) on all of the aforementioned pre-set multiple blocks, Electronic device.
  2. In Article 1, The above processor is, After acquiring the above FFT (Fast Fourier Transform) spectrum, the filter is configured to be designed based on block data corresponding to the above-mentioned multiple blocks, Electronic device.
  3. In Paragraph 2, The above block data includes the frequency and dominant direction of the above-mentioned multiple blocks, and The above processor is, Set to design a filter according to the following [Mathematical Formula 1], Electronic device. [Mathematical Formula 1] Here, is the angle of the above filter, and means the above main direction.
  4. In a method of operating a singularity enhancement electronic device through a noise image portion filter-in, i) A step of dividing first image data containing a fingerprint into a plurality of pre-set blocks; ii) a step of obtaining a Fast Fourier Transform (FFT) spectrum by applying a Discrete Time Fourier Transform (DFT) to one of the aforementioned divided, pre-set plurality of blocks; and iii) A step of outputting reconstructed image data by applying an inverse FFT after applying a designed filter to the above FFT spectrum; Includes, After the step of outputting the above restored image data, A step of repeating steps i), ii), and iii) across all of the aforementioned pre-set plurality of blocks; including, Method of operation of an electronic device.
  5. In Paragraph 4, After the step of acquiring the above FFT (Fast Fourier Transform) spectrum, the method comprises the step of designing the filter based on block data corresponding to the above-preset plurality of blocks. Method of operation of an electronic device.
  6. In Paragraph 5, The above block data includes the frequency and dominant direction of the above-mentioned multiple blocks, and The step of designing the above filter is, Step of designing a filter according to the following [Mathematical Formula 1]; including, Method of operation of an electronic device. [Mathematical Formula 1] Here, is the angle of the above filter, and means the above main direction.

Description

Electronic device for improving singularity through filter-in of a noisy image portion and method of operating the same The present invention relates to an electronic device for singularity enhancement through a noise image partial filter-in and a method of operation thereof, and more specifically, to an electronic device and a method of operation thereof that improves the accuracy of singularity recognition by dividing a fingerprint image into blocks, applying a filter to each block, and then removing noise. Fingerprint authentication is one of the widely used biometric technologies that requires image enhancement and feature extraction processes. These fingerprint authentication technologies can be primarily classified into correlation-based, minutia-based, and non-minutia-based methods. First, the correlation-based enhancement method is a technique that superimposes two images and calculates pixel-level correlations for different displacements and rotations. For example, there are methods to enhance fingerprint images, such as using adaptive filtering methods in the Fourier domain, using automated latent fingerprint recognition methods that represent correlation-based approaches, or using correlation-based enhancement methods through deep learning-based fingerprint liveness detection. Minutia-based enhancement methods are an important feature for comparing fingerprint images. For example, there are methods to emphasize the importance of fine extraction in benchmarks for fingerprint and distal phalangeal joint matching, or to improve fingerprint matching accuracy by using a hybrid approach that combines fine regions and directional fields. Non-minutia-based enhancement methods are alignment methods that use ridge shape, direction, and frequency images. For example, methods for removing noise from fingerprint images using convolutional neural networks, non-display-based approaches through deep learning-based fingerprint quality evaluation, and recent research have proposed new fingerprint enhancement techniques combined with deep learning. In addition, powerful fingerprint enhancement methods using deep neural networks, powerful minutia extractors by integrating deep networks with fingerprint domain knowledge, and deformable model-based approaches for verifying highly distorted fingerprints have been developed. This research has contributed significantly to overcoming the limitations of existing methods and achieving high accuracy even with low-quality fingerprint images. However, Minutia, one of these fingerprint recognition technologies, has the disadvantage of being difficult to detect in noisy and low-quality fingerprint images; consequently, inaccurate feature extraction and filtering can degrade matching performance even when using advanced algorithms. This project (result) is the result of the Local Government-University Cooperation-based Regional Innovation Project, conducted in 2024 with funding from the Ministry of Education and support from the National Research Foundation of Korea. (2021RIS-002) This research was supported by "Regional Innovation Strategy (RIS)" through the National Research Foundation of Korea (NRF) funded by the Ministry of Education (MOE). (2021RIS-002) FIG. 1 illustrates a block diagram of an electronic device according to various embodiments of the present invention. FIG. 2 is a flowchart illustrating how an electronic device operates according to various embodiments. FIG. 3 shows original image data containing fingerprints according to various embodiments. FIG. 4 shows image data in which a ridge-like region is identified in the original image and the fingerprint image is normalized according to various embodiments. FIG. 5 shows image data indicating the ridge direction calculated in each block of the fingerprint image important for applying an angle bandwidth allocation filter according to various embodiments. FIG. 6 shows image data representing the calculated ridge frequency of each block representing the ridge spacing used in the filtering process according to various embodiments. FIG. 7 shows an enhanced ridge image after applying an angle bandwidth allocation filter according to various embodiments. FIG. 8 shows a binarized and normalized image after filtering according to various embodiments. FIG. 9 shows a reliability image after filtering according to various embodiments. FIG. 10 shows a binary image according to various embodiments, representing a region where the mask value used for single point detection is 1 and the confidence level is 0.5 or higher. FIG. 11 shows an enhanced fingerprint image after applying the proposed angle bandwidth allocation filter according to various embodiments. FIG. 12 shows an image in which fine parts extracted from an enhanced fingerprint image important for fingerprint recognition, according to various embodiments, are more accurately extracted using the proposed method. Hereinafter, various embodiments of this document are desc