Research Demonstration

Research Work Illustrations

Work Title Goal/Application Method/Achievement Illustration
Thermal Face Recognition
  • To develop a reliable thermal face recognition system for national security applications;
  • Typical Applications: Prevention against terrorism at airports and borders, especially at nighttime;
  • Authentification to access a federal building or a computer-controlled device or facility.
  • A new orientation-based face recognition method is proposed, which utilizes Gabor wavelet transform and Hamming distance for identification;
  • The identification rate of the proposed method (called "Face Pattern Byte") achieves 97.92% with one modality image, and 100% with the score fusion of two modality images.
Night Vision Colorization
  • To enhance computer vision and human vision by multispectral image fusion and night vision colorization;
  • Applications: Performance improvement of night-time operation, surveillance and navigation, target recognition, and situational awareness;
  • Enhancement of national security.
  • The image fusion is based on the advanced DWT transform, and further optimized with regard to an IQ metric via an iterative DWT procedure;
  • The colorization is accomplished by mapping the color statistics of each segment in the multispectral (i.e., night vision) imagery to that of the auto-selected daylight picture;
  • A channel-based color fusion method is proposed for real time applications;
  • A quantative metric is proposed to measure and compare the image quality of colorized images.
Breast Cancer Detection
  • To detect breast cancers using digitized mammograms at their early stages;
  • The lifetime risk of developing breast cancer among American women is 13.4% (one in seven), exceeded only by lung cancer;
  • With digital mammograms computer-aided detection (CAD) is very useful to radiologists (as "a second pair of eyes").
  • A new breast cancer detection algorithm utilizing Gabor features is proposed, which involves in preprocessing, segmentation (generating alarm segments), and classification (reducing false alarms);
  • The proposed algorithm achieved TP (true positive rate) = 90% at FPI (false positives per image) = 1.21 in mass detection; and TP = 93% at FPI = 1.19 in calcification detection.
Bioinspired Color Enhancement
  • To simulate the contrast enhancement of center/surround networks and opponent analysis on the human retina;
  • The human retina ahs two center/surround layers, bipolar/horizontal and ganglion/amacrine; and four color opponents, red (R), green (G), blue (B), and yellow (Y);
  • The bio-inspired model is a powerful tool for image analysis and pattern recognition.
  • In our proposed enhancement model, the surrounding information is obtained using weighted average of neighborhood; excited or inhibited can be implemented with pixel intensity increase or decrease according to a linear or nonlinear response; and center/surround excitations are decided by comparing their intensities.
  • A difference of Gaussian (DOG) model is used to simulate the ganglion differential response.
  • The image quality of enhanced results is overall satisfied.
Glaucoma Detection
  • To detect glaucoma in its earliest stages by applying shape-based analysis techniques of retinal nerve fiber layer (RNFL) thickness to GDx-VCC (variable corneal and lens compensator) polarimetry data.
  • Wavelet-based (wavelet-Fourier analysis [WFA]), Fourier-based (fast Fourier analysis [FFA]), as well as the standard metric nerve fiber indicator (NFI), and all were compared as a function of disease stage;
  • Classification performance of WFA (ROC = 0.978; recently proposed method) was significantly better than FFA (ROC = 0.938) and NFI (ROC = 0.900).