Machine vision refers to the observation, collection, processing, and understanding of information from spatial measurements. Both observational and inferential data are required in order to produce results that are meaningful and useful for humans. The degree or power to which the inference engine is able to render the information understandable depends heavily on the use of robust algorithms and innovative architectures to perform automatic vision processing. In this paper, an expert system that incorporates such algorithms and architectures is described. The system, called El, is designed for imaging tasks using a personal computer (PC). The knowledge base for this expert system is a live or stored image, or sequence of images. The inference engine consists of a set of software tools--algorithms and paradigms that provide for spectral, regional, edge, and other types of analysis. These tools are implemented in a high-level, transportable language. The basic tool library can thus be continually expanded with system and user developments and routines to increase its versatility and utility for specific applications. Several industrial examples involving color, region, edge, and change detection are presented that use this system and illustrate its capabilities. The significance of this work lies in its attempt to package the basic processes of machine imaging in a user friendly, low-cost system so that such processing can be placed within the means of a far greater number of users than available today.