The enhanced Hoshen-Kopelman (EHK) algorithm for the analysis of connected components in images that are unbounded in one of their dimensions is introduced. The algorithm characterizes in a single pass the shapes of all the connected components in a multiple class image by computing the spatial moments, the area, the boundary, and the bounding boxes of the connected components. The algorithm is applied to a real-time surface defect simulation and to a Landsat image analysis. The algorithm's performance is compared to the performances of related algorithms.

Up to 1992, the international 5ESS telephone switch software update system, which maintains and updates released products software, was able to process the load of software update packages consistent with customers' needs. In May 1992, the introduction of a new international release, coupled with increased diversity of customer needs, amplified the complexity inherent to the management of large software projects, thus significantly increasing the demand on the software update delivery system. The usual result of such a situation is an increase in the package queue length and time. To bring the throughput to the required level, a solution was developed based on a theoretical model where two levels of abstraction of directed and undirected graphs were applied to represent the software update system. Using this graph model, a new tool was rapidly developed for the analysis and production of software update packages. Two weeks after the tool was released and over a period of three months, queue time for software update packages processing declined from 40 days to a steady state of 2–3 days. Key to the development of the technique was the intimate knowledge of the software development and update processes by a team of developers who supported the implementation of the technique. © 1998 John Wiley & Sons, Ltd.