A microcomputer based combined machine vision and expert system for irregular object classification

Abstract

In recent years, much work has been reported on the development of microcomputer based machine vision systems. A substantial portion of this research assumes most of the items subjected for machine vision inspection, can be categorized by a regular geometrical shape. Most of the vision systems, and microcomputer-based vision systems in particular, are designed to perform singular tasks. They are niche oriented and designed to be used in an inflexible environment, and are designed to process regular objects. These systems are prohibitively expensive for small industrial concerns. The objective of the research was to develop and design a low cost, microcomputer based, machine vision system to process irregular objects. The specific irregular object application for this research is the grading operations for the tree seedlings at the Iowa Conservation Commission Nursery in Ames. The main emphasis of the project has been on achieving optimum processing speed, and functional versatility. Various phases required to process the tree seedling were formulated and developed. The system made an extensive use of highly efficient run-length coding techniques, featuring comprehensive line segment tracking with flexible segment discrimination characteristics. Specific areas of the image were subjected to a set of logical masks, in order to isolate, identify, and measure, various parameters of the seedling. The acquisition and processing time was fraction of a second. It was concluded that having a high resolution and noise tolerant image capture system was essential for processing irregular objects like tree seedlings.