Research On Spray Control To Aim Toward Crop-Rows System Based On Machine Vision

As a large agricultural country, China must improve the performance of agricultural equipment for the competition of economy and sustainable development.Rough spraying technology in traditional way hardly adapt to the modern agriculture in plant protection."The basic way for agriculture's development is mechanization",which said by chairman Mao fifty years ago,has a deeply instructive to us now. With the development of agriculture mechanization,precision and intelligence in all aspects of agriculture are the main way nowadays.As traditional spraying not only leads to residue pesticide,but also waste pesticide,pollute environment, harm people, which is inevitably violate the request of modem agriculture.According to the study of home and aboard,given the pesticide spray precisely,the dosage of pesticide will be decreased by 40 to 60 percent compared with that of traditional way.Therefore it is necessary to research on spraying precisely and autonomously in time.On one hand,the labor conditions can be improved;on the other hand,the efficiency of pesticide can be improved and environment be avoided pollution.Machine vision,a high precision technology,is widely used in agriculture for taking the place of man's eye.In this article a spray control system of field crops based on machine vision is developed.The spray controlling equipment was designed to aim at the crop-rows.At first,A CCD camera was mounted on a tractor in order to capture real images,and send them to computer for processing,obtain the center line of crop-rows by Hough transform,finally a nozzle mark was used to move toward the line of crop-rows by means of step motor.The characters of young field crops is researched in this article,as many field crops plant in lines and grow as block one by one in early states.In order to study conveniently,suppose that each crop row have the same distance, each crop row has not interleaving and there is no weed in the field. Under these conditions some researches were developed. The corresponding results were showed as following:(1)Segmenting crops from its background is the first step to obtain the center line of crops-row.The method of segmentation was studied in this article.A changed method of watershed algorithm is used to segment field crop from its background,which proved efficient and simple compared with the traditional watershed algorithm.With respect to its time-consuming,the factor 2G-R-B was used to segment the field crops from its background.(2)Obtaining the center line of crop-rows is the key step of spray control based on crop-rows.A method was put forward to detect crop-rows by means of Hough transform based on connected component labeling.Firstly field crops were segmented from its background by means of super green (2g-r-b) algorithm,then its binary image was obtained using the algorithm of OSTU,finally the centroid coordinates of each field crops were computed after labeling,through which the center line of the crop-row was regressed by Hough transform.(3)Another method was put forward to obtain the center line of crop-rows with respect to timing process.Firstly the image of field crops were captured by CCD,then RGB model of the field crops was transformed to that of HSI,the H value on HSI model of the field crops was extracted and its binary image was obtained by means of OSTU,finally the binary image was processed by morphology erode and the centre line of crop-rows was.regressed by Hough transform.(4)The equipment of spraying system is design to aim at crop-rows. Compared to many research of fixing nozzles to spray where there are crops or weed,this article put forward to move the nozzle mark to aim at the crop-rows.In the system gear and gear strip moving by step motor, on the gear strip there is a nozzle mark,through which the experiment is done and the results showed its efficiency.(5)To solve the problem of spraying aiming at many crop-rows with many nozzles,a new idea was put forward. Field crops images captured by CCD were sent to computer for detecting the nearest crop-row to the center point of image in the horizontal direction by means of EUJIRID distance rule,of which the nearest crop-row will be choose as target to spray control.(6)The relationship between world coordinate and image coordinate is deduced in order to aiming at the crop-row precisely.The inner and exterior parameters are labeled and the mathematics model of single camera computing distance is deduced.(7)A feed-back system based on machine vision is established to improve the robust and adaptive performance of the system.(8)The software of spray control system of field crops based on machine vision was programmed by VC++, which including the module of image capture, image preprocessed, the center line of crop-rows extracted and the step motor controlled.(9) The power system was design to supply for the step motor and computer. A battery of 12 volt connected with a pure sine wave inverter of Cotek, through which alternating current can be produced.