|Edwards, Richard - USDA-RETIRED|
|Mossman, Albert - USDA-RETIRED|
|Miller, Ernie - ASAE MEMBER|
Submitted to: Applied Engineering in Agriculture
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: October 1, 2001
Publication Date: January 1, 2002
Citation: Pearson, T.C., Edwards, R.H., Mossman, A.P., Wood, D.F., Yu, P.C., Miller, E.L. 2002. Insect egg counting by image analysis and laser cutting of egg-laden mass rearing oviposition pads. Applied Engineering in Agriculture. 18(1):129-135. Interpretive Summary: Methods were developed to count eggs that were laid on commercial rearing pads. It is desired to have accurate egg counts on the pads because if too many eggs are on one pad, than the insects will be underfed and of low quality. Conversely, too few eggs on a pad will result in wasted feed for the insects. The methods used to count eggs were based on obtaining a gray scale image of the pad and using the image data to predict the number of eggs on a pad. Results indicate that prediction of egg counts can be obtained within +/- 10% of the actual value. In addition to egg counting, a procedure was developed to cut the pads with an infrared laser. The combination of egg counting by image analysis and laser cutting shows the feasibility of automatically cutting pads into segments with optimal egg counts. In addition, the counting method will allow entomologists to quickly and accurately count eggs from laboratory or large scale experiments.
Technical Abstract: Variability of egg quantities laid on rearing trays during large scale production of sterile insects can cause economic losses due to overfeeding or underfeeding. In addition, quality control measures to monitor egg numbers are both tedious and laborious. Methods were developed to count pink bollworm egg numbers on oviposition pads using histogram features from gray scale images of the pads. Egg count prediction by regression analysis produced highly significant predictive model equations when individual histogram bin values, or cumulative histogram bins were used as independent variables. Models using cumulative histogram bin values appeared more robust as the mean squared error values were slightly lower. A method was also developed to cut pads with 2 watts of power from a CO2 laser at a cutting velocity of 40 mm/sec. The pad was cut from the side opposite of the eggs and the resulting uncut layer of eggs were sheared off by a compressed air stream. Automation of this methodology would allow a mass rearing facility to obtain oviposition pad segments with the desired number of eggs +/- 10%, increasing insect yield and/or quality and minimizing diet costs. In addition, the methodology will allow entomologists to quickly and accurately count eggs from laboratory or large scale experiments.