LINEBURG


<< Пред. стр.

страница 22
(всего 28)

ОГЛАВЛЕНИЕ

След. стр. >>

lege of Engineering,
Holownicki R, Doruchowski G, Godyn A; Swiechowski W.2000.Variation of Spray
Deposit and Loss w ith Air-jet Directions applied in Orchards. J. agric. Engng Res., 2000, 77
(2),129-136


Tiansheng Hong,
Южно-китайский сельскохозяйственный университет
Wanzhang Wang
Henan сельскохозяйственный университет

ЭКСПЕРИМЕНТАЛЬНОЕ ИЗУЧЕНИЕ РАСПРЕДЕЛЕНИЯ ПЕСТИЦИДА,
РАСПЫЛЕННОГО ИЗ ГИДРАВЛИЧЕСКОЙ НАСАДКИ,
С ЦЕЛЬЮ ПРОФИЛЬНОГО МОДЕЛИРОВАНИЯ РАСПЫЛА

Резюме

Опрыскивание пестицидами по-прежнему является наиболее эффективным и
экономичным способом борьбы с болезнями растений и насекомыми-вредителями при
выращивании фруктовых деревьев. Однако применение химикатов в фруктовых садах
рассматривалось, как загрязнение окружающей среды. Поэтому в течение многих лет
существовала тенденция снижать объем пестицидов, применяемых для опрыскивания
фруктовых деревьев. С целью удовлетворения требований современной защиты расте-
ний и все более ужесточающихся экологических стандартов, должна быть разработана
технология эффективного и безопасного опрыскивания пестицидами. Опрыскиватель,
предназначенный для работы во фруктовых садах, должен обеспечивать соответст-
вующее осаждение химикатов на опрыскиваемой поверхности и минимальные потери
веществ. Значительный вклад в решение этой задачи при совершенствовании техноло-
гии опрыскивания внесло использование ультразвукового датчика. При помощи этого
датчика оказалось возможным проводить так называемое селективное распыление. Це-
лью данной работы было изучение распределения пестицида, выпускаемого из гидрав-


182
ISBN 5-88890-034-6. Том 1.


лической насадки, чтобы осуществить наиболее эффективное осаждение в зависимости
от контурной модели основы аэрозоля на ультразвуковом датчике.
Были проанализированы траектории и осаждение капель пестицида с учетом
положений гидродинамической теории и механизма распыления гидравлической на-
садки. Была разработана опытная установка на основе однокристального микрокон-
троллера (модель АТ89С52) с целью изучения распределения раствора пестицида, вы-
пускаемого из гидравлической насадки по горизонтальной поверхности. Стенд включал
ультразвуковой датчик, электромагнитный вентиль, датчик расходомера, транспортер и
насос для опрыскивания. Установка регулировала и регистрировала параметры опры-
скивания, такие как давление и объем распыла, скорость движения, а также собирала
осаждаемый раствор на специальном столе.
Испытания проводились с использованием двух насадок (форсунок) в виде по-
лого конуса (модель: 1/4MKB80200BCV-RW с VMD 130 ? m и 210 ? m
1/4MKB80320BCV-RW, производства фирмы Ikecuchi, Япония, в дальнейшем обозна-
чаемых 200 и 320) для изучения распределения раствора. Было установлено влияние
давления распыла и скорости движения на распределение раствора. Результаты опыта
показали, что скорость движения и давление оказывают значительное влияние на рас-
пределение пестицида. С повышением давления среднее расстояние осаждения увели-
чивается прямо пропорционально. При росте давления с 0.67 до 2,38 MPa среднее рас-
стояние осаждения из двух выбранных насадок увеличивалось на 464 мм и 391 мм, со-
ответственно. С повышением скорости движения опрыскивателя, среднее расстояние
осаждения из двух насадок снижалось, а когда скорость движения достигала опреде-
ленного уровня, среднее расстояние осаждения снижалось значительно. Повышение
скорости от 0,51 до 4,0 км/час привело к снижению среднего расстояния осаждения из
насадки 200 и 320 снижается на 251 мм и 173 мм, соответственно. Для дальнейшего
компьютерного расчета распределения раствора пестицида на основе опытных данных
была построена адаптированная логарифмически нормальная функция распределения.
По результатам данного опыта был определен принцип выбора параметров рас-
пыления и регулирования расстояния от насадки до кроны дерева.

Получено 16.05.2005.



Xiwen Luo1; Xuecheng Zhou1 ; Xiaolong Yan2
(xwluo@scau.edu.cn) (xcem@3126.com) (xlyan@scau.edu.cn)
1 2
( College of Engineering, Root Biology Center, South China Agricultural University,
Guangzhou 510642, P. R. China)

SEGMENTING ALGORITHM FOR MSCT IMAGES OF PLANT ROOT SYSTEM
BASED ON SPATIAL GEOMETRICAL FEATURES

Abstract. The root system is an important plant organ for nutrient and water uptake
from the surrounding medium. Due to its special growing environment and complex structure,
the root system is more difficult to quantify than the shoot. Observation and measurement of
root architecture in situ have been the technical bottleneck for research on plant root systems.
As an attempt to solve the above barrier, plant root images in situ were obtained in this study
using multi-slice spiral computed tomography (MSCT) imaging technology. In order to sepa-
183
ISBN 5-88890-033-8. Экология и сельскохозяйственная техника. СПб, 2005.


rate a root system completely from its surrounding medium, it is necessary to make effective
segmenting of the root images. Image segmentation, however, is very difficult in practice,
since most of the existing approaches to image segmentation are based on particular type of
images. In order to get a better segmenting effect, an interactive segmentation method was
developed based on the spatial geometrical features of the root system and the gray-level in-
formation of its MSCT images. After analyzing the gray-level histograms of the whole 3D
image and slice images, the threshold values were determined, and the binarization of root
images was processed. The images were processed using proper mathematical method accord-
ing to the feature of root 3D morphology. Finally special segmenting algorithm based on the
geometrical characteristics of root system was developed to remove most of the impurity
whose density is very close to root system. Special algorithm was designed based on the fol-
lowing principles: both the spatial location of root system between adjacent slices and the
change of its sectional shape and area are continuous. Because the rhizome in the CT images
near medium surface is very clear and easy to be recognized, the first step of the algorithm is
to manually locate and segment the rhizome regions in the 3˜5 slices of CT images near the
medium surface, so that the basement of the recurrence approach is setup. Then according to
the spatial continuity of the root system, the regions of root system downwards were revealed
slice by slice and the impure pixels were cleared. The results indicated that integrated algo-
rithms is an more effective method for segmenting the images.
Keywords Image segmentation, MSCT images, Plant root, Geometrical feature
Introduction. Soil is a barrier to direct visualization and measurement of root system
while it provides water and nutrients to plants. To clearly understand the process of root
growth and distribution belowground, various approached to observing roots have historically
been attempted, inlcuding the soil volume method in plastic tube, the mesh bag method, the
three-dimensional coordinate container method, the rhizotron technology and the isotope trace
method so on (Mao, 1994; Cheng et al., 1999). The present methods used in observing root
generally include two steps: sampling from soil and measurement. Most of sampling proc-
esses are destructive to the root system because it is necessary to dig and separate roots from
soil. Such sampling may not only break some small and thin roots which are required for root
morphological research, but also time-consuming and laborious. What is more, it is very diffi-
cult to keep root system without shift and deformation during sampling or later. Furthermore,
after sampling many traditional measurements are also taken manually. With the advancement
of relevant technologies, there are a few of new techniques for nondestructive sampling and
automatic measurement such as the rhizotron technology, the isotope trace method and com-
puter measurement methods based on digital image processing (Luo et al.,1999; Kimura et al.,
1999) . Most of them, however, can only provide some limited information or 2D (plane) data
in situ, but not intact and exact 3D morphological data. In fact, the lack of exact and facile
methods for observation and measurement in situ has always been the technical bottleneck for
research on plant root system (MacFall, 1998; Guo et al., 2001). As an attempt to solve the
above barrier, plant root images in situ were obtained by using the multi-slice spiral computed
tomography (MSCT) imaging technology, and the visualization of plant roots growing in dif-
ferent media was implemented after image processing and 3D reconstruction(Luo et al.,
2004). However, it is very difficult to separate a root system completely from its surrounding
medium using the available medical software developed for human organs imaging. There-
fore, it is necessary to develop special methods of image segmenting and 3D reconstruction
for plant roots so as to obtain clearer 3D root images (Luo et al., 2004). Image segmentation is
very difficult in practice, however, since most of existing approaches of image segmentation
are based on the particular type of images. In order to get a better segmenting effect, an inter-
184
ISBN 5-88890-034-6. Том 1.


active segmentation method will be developed based on the spatial geometrical feature of root
system and gray-level information of its MSCT images.

1 MATERIALS AND METHODS
The objective of this paper is to quantify the root system of a banyan (Ficus religiosa
L.) seeding grown in the soil. The raw images in situ of which were obtained by a scanning
equipment used for medical imaging (Sensation 4 MSCT, Siemens, Germany). The main
scanning parameters used were tube voltage of 120 kV, milliampere setting of 90 mAs, 1mm
slice thickness and 1mm screw pitch. After scanning, each sample yielded more than 200
slices of CT images with each slice consisting of an image matrix with 512 pixels X 512 pix-
els (Figure 1). These images of the root system in situ were preprocessed appropriately with
filtering, interpolation, and encapsulation. After that, these images were used as the raw data
for research on segmenting algorithm. In order to get a better segmenting effect, 3D threshold
segmenting for CT images were done according to the similarity of gray-level between slices
and within a slice for root system. Then utilizing the 3D morphological feature of root system,
region trimming was implemented with appropriate mathematical and morphological meth-
ods. Based on the spatial geometrical features of the root system, the impure pixels whose
density is close to the root system were wiped off from the CT images. Finally, the character-
istic regions were extracted and filled using seed fill algorithm.




(a) (b
)
Figure 1. MSCT Imaging of Plant Root System in Situ:
(a) slice of root section image (b) over 200 slices of CT images from a
root system

2 RESULTS AND DISCUSSION
2.1 3D threshold segmenting
In the segmentation method, the threshold values were determined after analyzing the
gray-level histograms of the whole 3D image and slice images, and the binarization of root
images was processed. In order to illustrate the whole distribution of the gray levels of the
images, the whole histogram of all images and the histograms of some typical images were
drawn. Figure 2 is the whole histogram, and the histogram of the slice 104 ( middle slice )
was shown in figure 3. In fact, the histograms of the slice 30 60 … 180 and the top slice
besides the middle slice were selected.



185
ISBN 5-88890-033-8. Экология и сельскохозяйственная техника. СПб, 2005.




1400000


1200000


1000000


800000


600000


400000


200000


0
00


00


00


00


00


00


00


01


01


12


38


85


94
0


0


0


0


0
20


40


60


80

10


12


14


16


18


20


22


24


26


28


30


32


35
Figure 2. Whole histogram for all images


7000


6000


5000


4000


3000


2000


1000


0
77




09


02


83


86


97
0




4


1


8


5


2


9


6


3


0


7


4
15


23


30


38


<< Пред. стр.

страница 22
(всего 28)

ОГЛАВЛЕНИЕ

След. стр. >>

Copyright © Design by: Sunlight webdesign