A framework for automated micro-manipulation of microscopic objects using visual information is developed and applied to diverse problems in biology and material science. These applications include microdissection of stretched and immobilized DNA molecules using an optical microscope and dynamic studies of crystal structure formation observed with a high power transmission electron microscope. Manipulation in this context refers not only to mechanical positioning or dissection of a specimen, but also to the variation of environmental parameters that elicit a response in the specimen. The common thread between these applications is a collection of computational techniques for image compression, autofocusing, object detection, tracking, and servo loop control. A second feature of our system is a distributed software architecture that allows the proposed dynamic experiments to be conducted remotely over the wide area network. In this context, the visual information processing is used to compensate for dynamic changes and the latencies in the wide area network. This system is now available over the global Internet.

In this paper, we present a system for detection and tracking of tubular molecules in images. The automatic detection and characterization of the shape, location, and motion of these molecules can enable new laboratory protocols in several scientific disciplines. The uniqueness of the proposed system is twofold: At the macro level, the novelty of the system lies in the integration of object localization and tracking using geometric properties; at the micro level, in the use of high and low level constraints to model the detection and tracking subsystem. The underlying philosophy for object detection is to extract perceptually significant features from the pixel level image, and then use these high level cues to refine the precise boundaries. In the case of tubular molecules, the perceptually significant features are antiparallel line segments or, equivalently, their axis of symmetries. The axis of symmetry infers a coarse description of the object in terms of a bounding polygon. The polygon then provides the necessary boundary condition for the refinement process, which is based on dynamic programming. For tracking the object in a time sequence of images, the refined contour is then projected onto each consecutive frame.>

We present a system for detection, tracking and representation of tubular objects in images. The uniqueness of the proposed system is twofold: at the macro level, the novelty of the system lies in the integration of object localization and tracking using geometric properties; at the micro level, is the use of high and low level constraints to model the detection and tracking subsystem. The underlying philosophy for object detection is to extract perceptually significant features from the pixel level image, and then use these high level cues to refine the precise boundaries. In the case of tubular objects, the perceptually significant features are anti-parallel line segments or, equivalently, their axis of symmetries. The axis of symmetry infers a coarse description of the object in terms of a bounding polygon. The polygon then provides the necessary boundary condition for the refinement process, which is based on dynamic programming. For tracking the object in a time sequence of images, the refined contour is then projected onto each consecutive frame. In addition, the system provides an axis of symmetry representation of object for subsequent scientific analysis.>