Abstract

IMAGE PROCESSING FOR ANALYSING SPRAY DEPOSITS. Procedure description In the present studies, digital image processing equipment was used to make spray pattern analysis. A fluorescent tracer was mixed in the spray liquid to identify the spray deposit. The tracer, illuminated with UV-light in the analysis, makes the droplets white while the rest of the leaf becomes black. Two different approaches have been made to document leaves. One approach is to photograph them with a normal 35 mm camera and make the analysis from the negatives. The other approach is to collect the images on a videotape. To obtain good results of analyses when using negatives the image must be illuminated uniformly. This causes more work before an image is ready for analysis than when using video images. In order to know the real size of a deposit, it must be related to something with known size. A ruler has therefore been put in the image for calibration. Small area deposits will give a lower contrast than larger areas. The reason for this is unknown, but one theory is that small areas contain less tracer in proportion to their area than larger areas. Therefore small areas give poorer contrast than big areas. It often happens that deposits are connected to each other and are thus segmented as one object. This is observed in the result when a small number of deposits is in the biggest interval which makes up 90-95 % of the total area. The reason for this is that the threshold is too high. This problem can be avoided by filtering the image. It requires that you can work with two fullsize images to avoid making the resolution worse. DEPOSIT DETERMINATIONS IN HORIZONTAL AND VERTICAL DIRECTION IN A DENSE PLANT STAND WITH THE IMAGE ANALYSER AND THE FLUORIMETER METHOD A preliminary investigation was made to investigate the possibility of minimizing the sampling procedure for deposit determination in field- and laboratory trials. Sample collection and analysis take up a great part of the time and work required. It is therefore important that this part of the research work can be rationalised with a preserved or even improved accuracy in the analytical values. A suitable image analyser was procured during the spring of 1987 and interest was focussed on the relationship between coverage (obtained with the image analyser) and the deposition as per cent of spray per hectare (obtained with fluorimeter). Spraying was carried out in a dense wheat crop after earing. Samples from nine different positions in the plant stand were collected. Deposited spray in per cent of amount of spray per hectare (fluorimeter) was chosen as independent variable and coverage in per cent (image analyser) as dependent variable. The values were used to fit a linear regression model. It was found that the linear curve (Y=0.734+1.538*X, Y="coverage (%)" and X="recovery (%)") would give the best correlation. The deposition studies revealed large differences in vertical direction and clearly emphasize the difficulty for droplets to penetrate dense crops. In addition, they also illustrate the very poor lateral spray distribution. Spraying with conventional nozzles in dense crops has a restricted possibility to give optimal results for pesticide control at deeper levels in the stand.

Keywords

image analyser; spray deposits; spray coverage; tracer; image processing; qualitative deposits; quantitative deposits; sprayer

Published in

Rapport - Sveriges lantbruksuniversitet, Institutionen för lantbruksteknik
1988,
ISBN: 91-576-3382-7
Publisher: Institutionen för lantbruksteknik, Sveriges lantbruksuniversitet

Permanent link to this page (URI)

https://res.slu.se/id/publ/125676