Dr. Hardie's research includes a wide range of topics in digital signal and image processing.  His focus is in the area of signal and image restoration and enhancement.  This includes developing innovative algorithms for overcoming limitations of non-ideal sensors through post-processing.  This work includes nonlinear filter design, multi-frame super-resolution, detector non-uniformity correction, and noise reduction.   He is also active in the areas of hyperspectral and medical image processing.  Hyperspectral image processing work includes anomaly detection, change detection, and image fusion.  The medical image processing research has focused on the detection and segmentation of pulmonary nodules from computed tomography images and chest radiographs.

Research Focus Areas and Projects

  • Super-resolution enhancement of digital video
    • This work seeks to produce improved resolution images with reduced aliasing by exploiting motion between frames in digital video.  Dr. Hardie is a leader in super-resolution research.  Along with several collaborators, Dr. Hardie received the Rudolf Kingslake Medal and Prize from SPIE in 1998 for work in this area. 

  • Signal and image restoration
    • Dr. Hardie has developed a number of novel algorithms to address the problem of signal and image restoration for both stationary and non-stationary signals, and Gaussian and non-Gaussian noise. 

  • Medical image processing

    • This work has focused on developing computer aided detection and segmentation algorithms for lung nodules in computed tomography (CT) imagery and chest radiographs to aid in the fight against lung cancer.  We are also developing computer aided diagnosis tools for automatically classifying nodules as benign or malignant.   
    • We are also now working on skin lesion segmentation and classification for computer aided skin cancer detection.

  • Detector non-uniformity correction for focal plane array imaging sensors

    • Developing scene-based detector non-uniformity correction algorithms for infrared focal plane arrays to treat the nonuniformity in the photoresponses of the individual detectors.  Non-uniformity correction reduces "fixed-pattern" noise in imaging sensors.

  • Atmospheric turbulence mitigation

    • We are developing algorithms for the mitigation of geometric distortion and blurring due to atmospheric optical turbulence.    
    • New simulation tools have recently been developed to generate realistic and accurate optical turbulence degradations for extended scenes under anisoplanatic conditions.  The simulation tools allow one to test and optimize turbulence mitigation algorithms quantitatively.

  • Hyperspectral image processing

    • We have developed novel image fusion methods for fusing low spatial-resolution hyperspectral imagery with high spatial-resolution broadband images.  
    • Anomaly and change detection in hyperspectral data.

  • Speech processing

    • Wideband speech regeneration from narrowband for improved speech quality through bandlimited channels.

  • Optical phased arrays

    • Nonmechanical beam steering for panning an imaging system using nematic liquid-crystal optical-phased-array (LCOPA).
  • Ceramic matrix composite microstructure image analysis

    • Ceramic fiber segmentation and tracking.  Anomaly detection of microstructural defects in continuous fiber reinforced composites.