Computer vision system for on-line sorting of pot plants using an artificial neural network classifier

Computers and Electronics in AgriculturePublished 1 May 1996

A.J.M. Timmermans, A.A. Hulzebosch

DOI10.1016/0168-1699(95)00056-9

Source URL

Citations62

SJR quartileQ1

SJR score1.83

SNIP2.35

TL;DR

A flexible grading system for pot plants and a neural network (NN) classifier are used as techniques for pattern recognition, which gave at least equal results as linear discriminant analysis (LDA) or quadratic discriminantAnalysis (QDA).

Abstract

A flexible grading system for pot plants is described. The system consists of a colour camera, an image processing system and specially developed software. It can be applied to several types of pot plants because of its implementation of learning techniques. Experiments are described for classification of a flowered plant and a cactus plant. Statistical discriminant analysis and a neural network (NN) classifier are used as techniques for pattern recognition. NNs gave for less complex applications at least equal results as linear discriminant analysis (LDA) or quadratic discriminant analysis (QDA). For more complex applications the NN classifier showed better results. Complexity for the described classification tasks can be associated with the distribution and shape of groups in the multi-dimensional feature space.

Keywords

ChemistryAgricultural and Biological Sciences

Elsevier eBooksIntroduction to Statistical Pattern Recognition

11,121 Citations1990Keinosuke Fukunaga

A system to determine the sex of fish from their length, as accurately as possible, using a set of labelled data, together with the class label (male or female), as a training set.

WORLD SCIENTIFIC eBooksHandbook of Pattern Recognition and Computer Vision

970 Citations2005C. H. Chen, L. F. Pau

European Journal of Operational ResearchIntroduction to computation and learning in artificial neural networks

124 Citations1990Egill Másson, Yih-Jeou Wang

A short historical overview of the development of artificial neural networks is given and some of the concepts and terms used within the field are introduced and 5 network models are described: The perceptron, the Hopfield network, theHopfield-Tank net, the Boltzmann machine and the Kohonen self-organizing network.

TechnometricsComputations for Variable Selection in Discriminant Analysis

93 Citations1975George P. McCabe

Pattern Recognition LettersA multiclass nonparametric partitioning algorithm

46 Citations1986Jan Talmon

This paper reviews a few algorithms from the literature which formed the basis for a procedure with the following properties: it generates a binary decision tree, it can handle mixed data types, it performs the feature and criterion selection simultaneously, and it is interactive and hence enables the user to control the design procedure.

TechnometricsComputations for Variable Selection in Discriminant Analysis

33 Citations1975George P. McCabe

Pattern RecognitionPattern classification using an efficient KNNR

30 Citations1992Saeid Belkasim, M. Shridhar +1 more

No preprocessing and approximation are involved in this technique which makes it suitable for overlapped classes, and the number of distance computations is reduced considerably without any increase in the error rate.

Pattern Recognition LettersComparison of a neural network and a piecewise linear classifier

16 Citations1991Z.-P. Lo, B. Bavarian

A neural algorithm proposed by Kohonen is used to design a neural classifier that is slightly better than the piecewise linear classifier, however the training algorithm is much simplier and easier to implement.

IFAC Proceedings VolumesArtificial Neural Networks for Plant Classification with Image Processing

3 Citations1995Thomas Rath

Building classification models with backpropagation networks gives best results, but in the light of generalization the classification accuracy was not sufficient and artificial neural networks have to be used in combination with knowledge-based systems.