Detecting Research Of Soybean's Plant Nitrogen Based On Image Processing Technology

In recent years, soybean production has greatly been developed and soybeanshave become the fourth largest crop inferior to rice, wheat, maize. However, everyyear there are a lot of soybean, soya-bean cake and edible oil which are importedbecause of developing animal husbandry, increasing population and insufficientedible oil. Nitrogen is important and indispensable element to soybeans. First of all,some essential nutrition elements are demanded in the course of the growth ofsoybean such as C , H , O , N , P , K , Ca , Mg ,etc., among which the demands forN , P , K are high. Soybeans need nitrogen 2-3 times higher than the rice and maize.It is reported that to produce 100 soybean grains, 7.2 kilogram nitrogen are neededto be absorbed. It is obvious that soybeans are such crop as need much nitrogen.Secondly, the protein in the soybean seed account for about 38% dry weight of theseeds, and the content of protein in the seeds is one of the major indicators whichweights the quality of soybeans. Moreover, at each growth stage in the crop theinternal material supersession of the crop is needed at the request of biologicalcharacteristic. This kind of material supersession is based on carbon supersessionand the center is nitrogen supersession. If carbon supersession and nitrogensupersession are matching, it will do good to strengthening root absorbing andpromoting the ability of 共合作用, which establish the material base for laterconserving type supersession. Otherwise, the disturbance between carbonsupersession and nitrogen supersession influence not only on the expanding typesupersession of crops but also on the accumulation of later conserving typesupersession. So it is particularly important to detect the soybean nitrogen.This paper combines the project of the national high-tech plan (863) "theresearch on nutrition information of the crop and shape parameter measuringtechnology"(2003AA209012), and a system for detecting the nitrogen of plantsoybean is designed by using computer vision technology. In this paper, applyingcomputer vision technology to detect the nitrogen of soybean plant can avoid theerrors arising from human diagnoses and possess accurate characteristic, which canfully reflect the role of computer in precision agriculture.Firstly, present research status of image processing technology is discussed,and the composition of the detecting system is introduced. The detecting system ismade up of hardware system and software system. Hardware system includescomputer and CCD scanner. Software is compiled under Visual C++ 6.0 by theauthor. In this research, the author adopts the scanner as the image sensor, whichreplaces CCD lens adopted in traditional research. The use of the scanner makesthe system high-resolution ratio, fine color degree and simple operating andsystematical forming. Meanwhile, it is also a innovation in this paper. Secondly, through analyzing the image characteristic of the plant and choosingthe method, the color and texture characteristic of soybean blade are abstracted.Since the symptom of lacking nutrition elements is mainly expressed by the colorand texture characteristic of plant blade, the abstraction of the blade color andtexture characteristic is an important content in this paper. The abstraction iscompleted by self-built detecting system and self-developed analyzing softwareprocedure. After the soybean blades are preprocessed, nine groups of characteristicvalues (R,G,B,H,I,S,L*,a*,b*) are obtained. Five groups of characteristicvalues are also obtained by a series of steps, namely catching image, cutting image,preprocessing image and transforming image into gray degree image. Thecharacteristic values include energy value, inertia distance, entropy value,correlativity value and part stationary value. According to the characteristic that the symptom of lacking nutrition elementsis mainly embodied by the blade color and blade texture, fourteen characteristicvalues abstracted from blade color and blade texture reflect the nutritioninformation of the soybeans in varying degree. If the statistical analysis method isused to study this subject with so many variables, the number of these variableswill increase the subject complexity. So the principal component analysis method(PCA) is applied to set up new variables as less as possible, and the new variablesare independent between every two new variables. Through analyzing concretePCA, for principal component expression that can represent these fourteencharacteristic parameters, as follow: F1=0.334R + 0.347G + 0.183B + 0.027H + 0.342I + 0.311 S + 0.346 L* -0.306a*+0.331 b*+ 0.145NL –0.228GXJ –0.456SZ –0.289XGX +0.193JB F2=0.127R + 0.079G -0.083B + 0.058H + 0.079I + 0.178 S + 0.092 L* -0.024a*+0.148 b*-0.551NL + 0.228GXJ + 0.413SZ –0.134XGX -0.595JB F3=0.031R -0.077G -0.345B + 0.763H -0.080 I + 0.147 S -0.070 L* + 0.180a*+0.080b*+ 0.346NL + 0.078GXJ + 0.222SZ –0.203XGX + 0.063JB F4=0.117R -0.170G + 0.578B + 0.146H + 0.114 I -0.289 S + 0.036 L* + 0.296a*-0.215b*+ 0.006NL -0.311GXJ + 0.545SZ + 0.017XGX + 0.076JBThen, the mathematics method is adopted to set up identifying function, as follow:y? = 26.167 + 0.0745F1 ? 0.3893F2 + 0.136F3 + 0.0036F4 Finally the identifying function is verified by experiment. It is indicated that...