Delivering algorithms, processes, and tools to enable geometric and colorimetric calibration for a novel low parallax multi-camera 360 imaging system. Architecting an image processing system capable of delivering extremely high resolution 360 video in real time.

• Delivered algorithms for registering, segmenting, and classifying high precision 3D point clouds obtained via scanning profilometry, a completely new imaging domain for the company.• Developed methods for profilometer characterization, simulation, and calibration for a scanning system capable of maintaining 10-micron precision within the point clouds.

• Invented a system protected by three US patents for the capture, analysis, and delivery of identification photographs. Since its introduction, this system has generated over 40% of all new product revenue.• Developed custom algorithms and classifiers using investigatory tool sets while delivering high performance C++ libraries directly into products.• Created fully functioning C# based GUI reference applications to ensure effective, rapid commercialization, provide an interactive debug environment, and enable the creation, audit, and correction of ground truth data.• Established the roadmap for the transition from image classifiers based on licensed, classical methods to those based on Convolutional Neural Networks to eliminate more than $300K in annual licensing fees.• Created an experimental prototype GUI to optimize the impact of image curation algorithms on the workflow of the semi-automatic photo album creation. The concepts developed here were incorporated into products.

Researched in the fields of image retrieval, face recognition, pattern recognition, image understanding, relevance feedback, video processing.• Supervisor pro tempore for the Visual Information Management team.• Conducted research in face recognition in still images and video for both consumer and professional domains. This work provided the groundwork for technologies incorporated into products at Kodak Alaris.• Created a wide range of tools for feature extraction, analysis, modeling, clustering, retrieval, training and classification, data visualization, ground truth collection, and workflow analysis.• Created a high-performance face and eye tracking system for use in a 3D, immersive imaging research project. Provided faster and more robust tracking performance than commercially available systems at the time. • Invented methods for analyzing images to determine candidate regions for text placement as well as choosing appropriate accent colors for text and mats for image collages. • Delivered core functional components as commercial grade C++ libraries, and fully functioning GUI demonstration applications.• Researched unified feature frameworks that combine semantic features such as face detection and recognition with semantic region and image classifications with primitive features in a unified similarity framework driven by user interaction via relevance feedback.• Investigated feature-based indexation and retrieval for whole images and regions of interest for similarity-based retrieval and ranking, duplicate detection, and image matching.• Researched topics include shot boundary detection, key frame selection, video summarization, face (and eye) tracking, pupil and Purkinje reflection segmentation and tracking. Video processing libraries were implemented as Microsoft DirectShow filters with a C++ GUI.