Aquacultural Engineering 24 2000 33 – 57
Classification of rotifers with machine vision by shape moment invariants
Chan-Yun Yang, Jui-Jen Chou
Department of Agricultural Machinery Engineering, National Taiwan Uni6ersity, No.
136
Chou-Shan Road, Taipei,
106
Taiwan, ROC Received 3 April 2000; accepted 25 August 2000
Abstract
An automated system for the identification of rotifers under a microscope with machine vision by shape analysis has been developed, which tends to be substituted for human
appraisal. A suitable image recognition algorithm was proposed and the results were discussed in detail. In this study, rotifers were classified into the exact types despite the
debris, which appeared from sludge in the degraded water or from rotifer carcasses. Two stages of a discrimination model based on shape analysis were built: one was to separate
debris from rotifers, and the other was to classify rotifers into three groups. A set of shape descriptors, including geometry and moment features, was extracted from the images. The set
of shape descriptors had to satisfy the RST rotation, scaling, and translation invariance. Shape analysis was proved to be an effective approach since the classification accuracy was
approx. 92. The results from different classification approaches were also compared. The machine vision system with shape analysis and the 2-stage discrimination model had a
greater effect on the reduction of manpower requirement for the classification of rotifers. © 2000 Elsevier Science B.V. All rights reserved.
Keywords
:
Rotifer; Classification; Pattern recognition; Shape analysis www.elsevier.nllocateaqua-online
1. Introduction
Rotifer is an excellent first live feed for larval fish because of its small size, slow swimming speed, habit of staying suspended in water, high reproductive rate, and
Corresponding author. Tel.: + 886-2-23635375; fax: + 886-2-23627620. E-mail address
:
jjchouccms.ntu.edu.tw J.J. Chou. 0144-860900 - see front matter © 2000 Elsevier Science B.V. All rights reserved.
PII: S 0 1 4 4 - 8 6 0 9 0 0 0 0 0 6 5 - 0
possibility for mass production. Rotifer is also valuable for larval fish because of its amino acid composition and high digestibility Watanabe et al., 1983, it can be
enriched with fatty acids and antibiotics, and can be used to transfer these substances into fish larvae Lubzens et al., 1989.
Estimating the density of rotifers is a general method for assessing and monitor- ing the physiological condition of rotifer cultures Hoff and Snell, 1997. This
assessment depends upon the computation of the number of rotifers and carried eggs. Since rotifers are small, a sample from the culture tank needs to be examined
under a microscope, so that the number of rotifers as well as carried eggs can be counted one by one. Although this is important, the counting is time-consuming
and labor-intensive Fulks and Main, 1991. Further, errors might be introduced due to worker’s fatigue and carelessness. Hence, the development of an automatic
system to substitute for human counters is required.
Rotifers are similar in shape, except when bearing eggs. Rotifer size has been reported in the range of 123 – 292 mm in length and 114 – 199 mm in width Snell and
Carrillo, 1984. The body size of rotifers is primarily genetically determined which changes with environmental factors like temperature, salinity, or food type. Rotifers
have a repeatedly parthenogenetic life cycle that contains both asexual and sexual phases. The eggs produced from the sexual phase are slightly bigger in size and are
more spherical than those from the asexual phase with smooth oval appearance. Sexual eggs are covered with tough shells. The shells protect the eggs from extreme
environments Hoff and Snell, 1997.
The major objective of this study was to develop a system for the rotifer identification using pattern recognition under a microscope. Following the conven-
tional operation steps, a shape analysis technique is to be employed to characterize the difference between various types of rotifers and to be substituted for human
inspection. A set of features was introduced to make the identification depend only on rotifer’s shape despite its orientation, dimensions, and location. This study
concentrated on three types of rotifers: rotifers without egg, rotifers with one egg, and rotifers with two eggs. Functions on the developed system included identifying,
counting and then recording the number of rotifers which carried the eggs automatically.
2. Materials and methods