Avi Kak

Vision-based navigation by a mobile robot with obstacle avoidance using single-camera vision and ultrasonic sensing

This paper describes a vision-based navigation method in an indoor environment for an autonomous mobile robot which can avoid obstacles. In this method, the self-localization of the robot is achieved by a model-based vision system, and nonstop navigation is realized by a retroactive position correction system. Stationary obstacles are avoided with single-camera vision and moving obstacles are detected with ultrasonic sensors. We report on experiments in a hallway using the YAMABICO robot.

Content-based image retrieval from large medical databases

The main goal of content based image retrieval is to efficiently retrieve images that are visually similar to a query image. In this paper we focus on content based image retrieval from large medical databases, outline the problems specific to this area, and describe the recent advances in the field. We also present some of the more significant results obtained with ASSERT (Automatic Search and Selection Engine with Retrieval Tools), the content based image retrieval system developed in our laboratory.

Deforming virtual objects interactively in accordance with an elastic model

Abstract We show how interactive deformations of a virtual 3D object can be carried out by using a hierarchical implementation of the finite element method (FEM). Basing deformations on the concepts of elasticity gives the human a measure of predictability when deciding where to apply forces to the object so that a desired shape would ensue. As is well known, one of the most powerful tools for analysing elasticity is the FEM, but the computational burden associated with a straightforward application of FEM to the problem at hand would make it too slow for any interactive process on even the fastest workstations. We have therefore developed a method in which the computational burden of FEM is alleviated by carrying out the FEM analysis at two different resolutions; a coarse resolution for a 3D calculation of the deformations and, subsequently, a finer resolution for just the surface layers of the object for a better (and smoother) delineation of the object shape. For the case of analysing the surface layers using the finer resolution, we show how a plate-theory version of FEM can be employed.

An interactive framework for boundary delineation for medical CBIR

CBIR systems designed for medical applications often require that a human in the loop demarcate the pathology bearing regions in the image, since fully automatic extraction of such regions is still not possible. In our CBIR system for the domain of HRCT images of the lung, physicians do not find this interaction too onerous since the boundary surrounding a pathology bearing region does not have to be precise. But in our new domain, a liver image database, that is unfortunately not the case. The boundaries supplied by the physician must correspond precisely to the outline of the liver or to the boundary of a pathology bearing region inside the liver. To meet this demand, we have developed a new user interaction framework for semi-automatic boundary extraction. All that a physician has to do is to click on a couple of pixels on the boundary to be extracted. The system then tries its best to extend the boundary as far as possible, sometimes even extracting the entire contour correctly. When errors occur, all that the physician is called upon to do is to click on where a correction to the boundary needs to take place. In this manner, an entire boundary can be specified with very little input from the human, which is a most important consideration with physicians as he/she can hardly be expected to click on every boundary point.

Globally Optimal Interactive Boundary Extraction Using Markov Chain Modeling

We present a novel boundary-based (discontinuity tracking) hierarchical statistical criterion to address the interactive contour extraction problem. Our criterion relies on a Markov Chain representation of the boundary and can be efficiently optimized using Dijkstra’s algorithm for solving the shortest paths problem. Unlike other criteria optimized with Dijkstra’s algorithm, ours is capable of extracting geometrically complex boundaries even when the features incorporated in the objective function are based only on user markings on a small part of the image. The critical quantity in our criterion that yields the above-mentioned results is a normalization factor that boosts the probability of a particular boundary segment based on the candidate boundary segments in its vicinity. Although similar in spirit to the technique of non-maximum suppression routinely employed in edge detection, our method boosts gradually the probability of a particular segment given its surroundings using windows of increasing size in a hierarchical fashion.