Below are brief descriptions of current and previous projects. I will add more information when I get a chance.

Geographic Image Analysis Using Interest Points

This ongoing project investigates the application of interest point detectors and descriptors to analyzing geographic imagery such as satellite images or aerial photographs. We are exploring this in the context of content-based image retrieval and land-cover classification.


Terrestrial Remote Sensing for Monitoring Atmospheric Particulates

This recently iniated project investigates the use of image processing and computer vision for monitoring air pollution, specifically particulate matter. Both model and data driven approaches are being considered.


Measuring Respiratory Burst in Alveolar Macrophages

This ongoing project explores automated means for measuring respiratory burst in cultures of alveolar macrophages (white blood cells in the lungs). We are first considering standard approaches, such as watershed segmentation and the generalized Hough transform. This work is being done in collaboartion with biologists at UC Merced.


Spatial Analysis of Unstructured Proteins

This project is just starting. Unstructured proteins are receiving increased attention. However, very little work has been done on characterizing their "unstructuredness." We plan to investigate whether spatial analysis techniques can be applied to this problem. This work is being done in collaboration with biolgists at UC Merced. (Image courtesy of Mike Colvin.)


Using TEM to Analayze Nanoparticles Suspended in a Film

This project was done in collaboration with material scienctists at UC Merced. It involved using image processing and computer vision techniques to determine the spatial distribution of nanoparticles suspended in a polymer film. Rather than slice and image the cross-section of the film, we investigated whether the depth of the particles could be estimated through sets of tilted images.


Quantitative Analysis of Computer Simulations

This project was done at LLNL in the Sapphire Data Mining team in conjunction with physicists. Computer simulations of physical phenoma produce large output datasets. We investigated the use of image processing and computer vision techniques to provide a quantitative alternate to visual analysis of these outputs. (Image from UCRL-MI-126772 Rev. 2.)


Quantitative Comparison of Experimental and Simulation Data

This project was done at LLNL in the Sapphire Data Mining team in conjunction with physicists. It involved using image processing and computer vision techniques to compare experimental and simulation data of physical phenomena. (Image from UCRL-CONF-208568.)


Habitation Detection in Remote Sensed Imagery

This project was done at LLNL in the Sapphire Data Mining team. It explored using texture features to detect inhabited regions in high resolution, multispectral remote sensed imagery. (Image from UCRL-CONF-201981.)


Salient Points for Tracking Moving Objects in Video

This project was done at LLNL in the Sapphire Data Mining team. It explored salient points for tracking objects in video. (Image from UCRL-CONF-208738.)


Texture Analysis of Remote Sensed Imagery

This work was perfomed in the Vision Research Lab at UCSB. It investigated texture features for a range of remote sensed image analysis tasks, including content-based image retrieval, land-cover classification, texture motifs, and spatial relationships.


Mining Image Datasets Using Perceptual Association Rules

This work was perfomed in the Vision Research Lab at UCSB. Association rules were adapted to data mine spatial relationships in remote sensed imagery.


Analyzing Confocal Microscopy Images of Retinal Cells

This work was perfomed in the Vision Research Lab at UCSB in conjunction with biologists. It investigate image processing and computer vision techniques for analyzing confocal microscopy images of retinal cells at various stages of a retinal detachment disease. (Image from Retinal Cell Biology Laboratory, UCSB.)


Categorical Image Search on the WWW

This work was perfomed in the Vision Research Lab at UCSB. This project investigated image search on the world wide web. Image content (color and texture features) and associated text were used to search for visually and semantically relevant images. A categorical structure of the websites containing the images was used to improve the results.