Unusual vision - machine vision applications at the NCEA

Abstract

Purpose - The paper draws together a range of somewhat unusual machine vision applications, with an integrating overview. Design/methodology/approach - The National Centre for Engineering in Agriculture is amassing a portfolio of projects involving machine vision. These include identification of animal species, visual counting of macadamia nuts, analysis of animal behaviour and a number of quality control functions. DirectX tools have been developed and are applicable across a wide range of applications, while theory is extended in several ways. Findings - Most of the projects are still in progress, but results are reported on the degree of success of a range of methods. Strategies and algorithms are discussed. Research limitations/implications - Vision-based solutions are applied to a diversity of tasks. There will be a continuing stream of such problems with abundant opportunities for research. Practical implications - The projects are in essence practical, although they have inspired new methodologies. They are conducted in close collaboration with the industries involved and will be deemed to have failed if the outcome is not put to commercial use. Originality/value - The paper draws together a portfolio of projects, allowing an analysis to be made of the features that unite and differentiate them. It will be of interest to both researchers and those with instrumentation problems.