580 California St., Suite 400
San Francisco, CA, 94104
Key research themes
1. How can Gabor filters be optimized and adapted for enhanced feature extraction in biometric recognition systems?
This research area investigates the application and optimization of Gabor filter-based feature extraction techniques to improve biometric recognition accuracy and robustness. It matters because biometric modalities such as iris, face, and fingerprint recognition critically depend on effective feature representation that is invariant to rotation, scale, illumination, and noise, where Gabor filters provide biologically inspired texture and frequency-selective representations. Advances focus on multi-modal fusions, parameter tuning, and algorithmic integration with classifiers to address unimodal system limitations and enhance performance.
Key finding: This paper demonstrates the use of Gabor filters for simultaneous feature extraction from both face and fingerprint modalities within a multimodal biometric system, overcoming limitations of unimodal systems. The authors... Read more
Real-time Client Authentication in Cyberspace using the GLBPD and Three-stream 3D-CNNs with the LSTM
Key finding: The authors propose a real-time iris authentication framework combining Gabor filters with local binary patterns on diagonal planes (GLBPD), feeding sequential iris feature maps into three parallel 3D-CNNs followed by LSTM... Read more
Key finding: This study combines multi-wavelet features using discrete wavelet transform with Gabor features to capture complementary textural and localized iris patterns for SVM-based iris verification on the UBIRIS V1.0 dataset. The... Read more
Key finding: This paper presents a neural network-based face recognition system that uses Gabor filter coefficients at multiple scales and orientations to represent facial images, along with image morphing and RMS contrast scaling for... Read more
Key finding: The authors develop a hybrid genetic algorithm and neural network system for classifying grayscale images using multi-resolution stacks formed by Gabor filtered images at various scales and orientations. The genetic algorithm... Read more
2. What methodological advances enable the enhancement and optimization of Gabor filters for image denoising and texture segmentation?
This research area focuses on the design, optimization, and hardware implementation of Gabor filters to improve image denoising and texture segmentation quality, preserving salient features like edges and textures while mitigating noise. It matters due to the critical role of noise reduction and accurate texture analysis in varied image processing applications such as medical imaging and document analysis. This area includes algorithmic modifications, parameter tuning, and hardware accelerations for real-time, low-power operations.
Key finding: This work presents a low-power, portable hardware accelerator model implementing Gabor filter-based image denoising focused on edge detection and noise reduction in biomedical ultrasound, CT, and MRI images. The authors... Read more
Key finding: The study employs Gabor filters combined with convolutional neural networks (CNN) to classify mask usage positions in facial images. The Gabor filter extracts texture and edge information as preprocessed features that improve... Read more
Key finding: The paper proposes a biologically inspired multi-channel filtering scheme based on Gabor filter banks for separating text from non-text regions and script recognition in complex Indian bilingual document images. The authors... Read more
Key finding: This article introduces circular Gabor filters (CGF), a rotation-invariant modification of traditional Gabor filters designed for texture segmentation tasks requiring rotational invariance. The authors define the CGF... Read more
Key finding: This research combines the use of Gabor filters for texture analysis with particle swarm optimization (PSO) to enhance unsupervised retinal blood vessel segmentation in fundus images. Applying contrast enhancement (CLAHE) and... Read more
3. How can Gabor filter-based feature extraction be effectively integrated with machine learning models for disease diagnosis and medical image analysis?
This research stream explores the combination of Gabor filter feature extraction techniques with classical and modern machine learning models to enhance diagnostic accuracy in medical image analysis tasks such as cancer detection and retinal disease classification. It emphasizes the importance of extracting discriminative textural features from histopathological and radiological images and the subsequent classification performance improvements through optimized feature selection, fused descriptors, and improved learning algorithms.
Key finding: The authors apply Gabor filter-based statistical feature extraction on invasive ductal carcinoma histopathological images across varying sample sizes and wavenumbers, then train several machine learning classifiers including... Read more
Key finding: This study integrates Gabor filter features with gray-level co-occurrence matrix (GLCM) features, optimizing them through a weighted average gravitational search algorithm (WA-GSA), and applies them to breast mass... Read more
Key finding: Focusing on fingerprint image enhancement, this paper proposes a combined filtering approach integrating diffusion coherence filtering and 2D log-Gabor filtering to address limitations of traditional Gabor filters in... Read more