CN106872411A - A kind of tea tree tender leaf recognition methods based on ECG concentration differences - Google Patents

Abstract

The invention discloses a kind of tea tree tender leaf recognition methods based on ECG concentration differences, comprise the following steps：(1) reflectivity at its characteristic wave bands is obtained for fresh leaves of tea plant；(2) relational expression based on catechins ECG (hereinafter referred to as ECG) concentration in tealeaves Yu characteristic wave bands reflectivity, obtains the ECG concentration of the fresh leaves of tea plant；(3) the ECG concentration thresholds based on tea tree tender leaf, tender leaf identification is carried out to the fresh leaves of tea plant.A kind of tea tree tender leaf recognition methods based on the analysis of ECG concentration differences of the present invention, can be identified to the old leaf of tea tree and fresh leaf, and the degree of accuracy is high, strong antijamming capability, adaptability are good.

Description

A kind of tea tree tender leaf recognition methods based on ECG concentration differences

Technical field

The present invention relates to tea picking manufacture field, and in particular to a kind of recognition methods of tealeaves tender leaf.

Background technology

Tea picking is a work for quite taking manpower in Tea Production, typically to account for the 50% of tea garden management workload More than, and major part tea area of China is all based on the production of Famous High-quality Tea, and it is higher to pluck required precision, will typically reach a bud two The standard of leaf, such as making one kilogram of superfine Junshan Silver Needle Tea dry tea will pluck up to ten thousand bud heads.

As economic continues to develop, the large quantities of labours in rural area shift to secondary industry, the tertiary industry, and many tea areas occur Labor shortage problem.Other Famous High-quality Tea is with short production cycle, and " tealeaves is a division of day and night grass, and it three days is precious a, Chi Cai early to adopt in farmer's proverb road It is within three days root grass ", illustrate that tea picking is restricted to time very strong, it is difficult to concentrate enough labours, Er Qieyou in the short term In tea picking work, standard of plucking differs in picking process, can also reduce the quality of Famous High-quality Tea.

In order to solve the problems, such as that tea picking consumes a large amount of manpowers, people have attempted tea machineryization harvesting, but current Tea machineryization pluck there are problems that it is a lot, first be influence tealeaves quality because present main flow picking mechanical does not have The function of standby identification, using extensive machine cuts mode, has mixed many old leafs and old stalk and some weeds in tender leaf, and And many tender leafs can be broken during harvesting；Next to that influence tea tree grows, tea tree by after mechanical picking, Canopy lamina quantity is reduced, and leaf layer is thinning, and growth potential weakens speed and accelerates.Therefore the Mechaniaed harvest of cutting type is mainly used in The all relatively low large tea production of economic benefit and nutritive value.

And for the harvesting of Famous High-quality Tea intelligent machine, theory stage is substantially also in, it is no to be applied to reality The system of production.This is primarily limited to the CF feature that current tea picking position identifying system is based on tender leaf, no The production environment of complexity, such as different growth conditions of weather, the change of sunshine and tea tree can be perfectly suitable for.

The content of the invention

The invention provides a kind of tea tree tender leaf recognition methods based on ECG concentration significance analysis, can be to the old of tea tree Leaf and fresh leaf are identified, and the degree of accuracy is high, strong antijamming capability, adaptability are good.

A kind of tea tree tender leaf recognition methods based on ECG concentration differences, comprises the following steps：

(1) reflectivity at its characteristic wave bands is obtained for fresh leaves of tea plant；

(2) relational expression based on catechins ECG (hereinafter referred to as ECG) concentration in tealeaves Yu characteristic wave bands reflectivity, obtains Take the ECG concentration of the fresh leaves of tea plant；

(3) the ECG concentration thresholds based on tea tree tender leaf, tender leaf identification is carried out to the fresh leaves of tea plant.

Any external difference has its internal cause, and the tender leaf for distinguishing tea tree will not only enter with old leaf from mode of appearance Hand, more will in it chemical component difference start with, ECG as tealeaves major function inclusion, be to evaluate tea quality One of standard, the present invention can be visually using high light spectrum image-forming technology combination chemometrics method and image processing techniques Represent distribution maps of the ECG in tea tree, then realize that tea tree tender leaf is recognized, objectively improve tealeaves identifying system to different cultivars Adaptability, the enhancing of tea tree provide new thinking to the robustness of environment to develop famous green tea leaf intelligent identifying system.

Preferably, the fresh leaves of tea plant from tea tree breed chrysanthemum spring tea tree, the tea tree of Zhejiang agriculture 25 or to meet white tea tree.

Preferably, characteristic wave bands are 541nm, five wave bands of 616nm, 654nm, 677nm and 695nm.

Preferably, ECG concentration is with the relational expression of wave band reflectivity：

Y_{ECG concentration}=44.988-0.289 λ₅₄₁+4.205λ₆₁₆-6.82λ₆₅₄+3.632λ₆₇₇-0.77λ₆₉₅

In formula：λ_aRepresent the reflectivity at a nm wave bands, Y_{ECG concentration}Represent ECG concentration.

Preferably, being directed to different tea trees, the ECG concentration thresholds are：

Chrysanthemum spring tea tree, ECG concentration values >=79mg/g is the first leaf position blade；

White tea tree is met, ECG concentration values >=64mg/g is the first leaf position blade；

The tea tree of Zhejiang agriculture 25, ECG concentration values >=66mg/g is the first leaf position blade.

Preferably, step (1) described fresh leaves of tea plant is the blade of the 1-6 leaves position since the terminal bud of tree plant down, in Between be free of fish leaf.

Preferably, the reflectivity in step (2) at a nm wave bands is corrected by following black and white plate：

In formula：

λ₀It is the sample primary reflection spectrum obtained in hyper-spectral data gathering camera bellows, B is covering camera mirror under total darkness environment The demarcation reflectance spectrum (reflectivity is 0) of head collection, W is white standard reflection spectrum (reflectivity is 100%), and DN is multiplication factor (value is 4095).

Compared with prior art, the present invention has following beneficial technique effect：

(1) strong adaptability, the inventive method is notable by analyzing the ECG concentration differences of different varieties of tea plant various position leaves Property, for later use Machine Vision Recognition tender leaf provides the standard for generally adapting to, this standard will not be because of different cultivars tea Set the difference of tender leaf mode of appearance and can not recognize tender leaf；

(2) robustness is good, and the inventive method generates the concentration profile of ECG by extracting the characteristic wave bands of ECG, in tea place In, except containing ECG in tea tree, ECG is essentially free of in sky, soil and weeds, therefore the inventive method can be fine Ground shielding ambient interferences；

(3) degree of accuracy is high, method proposed by the present invention by tealeaves ECG differences distinguish old leaf and tender leaf, can be with The identification problem of advancing coloud nearside system biological information is solved, internal chemical composition information and mode of appearance information is comprehensively utilized, so as to improve The accuracy rate of tender leaf identification.

Brief description of the drawings

Fig. 1 is tea tree various position leaves ECG content difference figures；

Fig. 2 is chrysanthemum spring tea leaf ECG concentration distribution pcolor；

Fig. 3 is the R component figure of ECG concentration distribution pcolors in Fig. 2；

Fig. 4 is the G component maps of ECG concentration distribution pcolors in Fig. 2；

Fig. 5 is the B component figure of ECG concentration distribution pcolors in Fig. 2.

Specific embodiment

Embodiment

Chrysanthemum spring, Zhejiang agriculture 25 are gathered first, each 30 fresh leafs in the leaves of kind tea trees 6 position of frost 3 are met, altogether 3*6*30= 540, these tealeaves blades are divided into 54 samples by tea tree species and leaf position, i.e. 10 fresh leaf blades of each sample, used Hyperspectral imager (ImSpector V10E, Spectral Imaging Ltd., Oulu, Finland) scans the bloom of blade Spectrogram picture, after the end of scan, fresh leaf freeze-drying 24 hours is pulverized, using high performance liquid chromatograph (Shimadzu LC-20, Japan) the ECG concentration of each sample is measured, the ECG concentration value statisticses of blade are as shown in table 1.

Table 1ECG statisticses (J represents that chrysanthemum spring, Z represent that Zhejiang agriculture 25, Y is represented and meet frost)

ECG concentration differences by SPSS softwares respectively to the various position leaves of different varieties of tea plant carry out significance analysis, Analysis result is as shown in Table 2-4.

2 chrysanthemum spring of table various position leaves ECG concentration differences significance analysis result

Table 3 meets white various position leaves ECG concentration differences significance analysis result

The various position leaves ECG concentration difference significance analysis results of 4 Zhejiang agriculture of table 25

With reference to Fig. 1, in Fig. 1：Ordinate is ECG concentration (mg/g), and the J in abscissa represents that chrysanthemum spring tea tree, Z represent Zhejiang The tea tree of agriculture 25, Y are represented and are met white tea tree, and such as J-1 represents the upper first leaf position blade of chrysanthemum spring tea tree, and J-n represents chrysanthemum spring tea tree Upper n-th leaf position blade, other are similarly.

a₁、b₁、c₁、d₁、e₁The significant difference of ECG concentration between expression chrysanthemum spring various position leaves

(P<0.05)；

a₂、b₂、c₂、d₂、e₂、f₂Significant difference (the P of ECG concentration between the expression various position leaves of Zhejiang agriculture 25<0.05)；

a₃、b₃、c₃Significant difference (the P of ECG concentration between white various position leaves is met in expression<0.05).By significance analysis result It can be seen that, the ECG concentration values of the first leaf position blade of three kind tea trees and the ECG concentration value significances of difference of other leaves position blade It is 1 to be worth, and the level of signifiance is extremely notable.Therefore, the concentration value of ECG can be as the mark for distinguishing three kind tea tree tender leafs and old leaf It is accurate.

Based on tealeaves tender leaf ECG concentration thresholds, tender leaf identification is carried out, specific threshold value is：Chrysanthemum spring tea tree, ECG concentration values >=79mg/g is the first leaf position blade；White tea tree is met, ECG concentration values >=64mg/g is the first leaf position blade；The tea tree of Zhejiang agriculture 25, ECG concentration values >=66mg/g is the first leaf position blade.

Black and white plate correction is carried out to the original high spectrum image for gathering by formula (1), the softwares of ENVI 5.1 are then used Select the area-of-interest of high spectrum image and extract averaged spectrum.

Average reflectance spectra characteristic wave is obtained using the average ECG contents of blade and the fitting of average reflection spectrum characteristic wave band The relation of section and average ECG is shown below：

Y_{ECG concentration}=44.988-0.289 λ₅₄₁+4.205λ₆₁₆-6.82λ₆₅₄+3.632λ₆₇₇-0.77λ₆₉₅ (2)

Referring to Fig. 2, according to characteristic wave bands relative reflectance and the relation of ECG concentration, by each tealeaves blade in 5 features Relative reflectance image at wavelength is converted into ECG concentration distribution images.It is with chrysanthemum spring tea leaf ECG concentration distribution image Example.The first leaf position blade~the 6th leaf position blade is followed successively by figure from left to right.

R component figure, G component maps and the B component figure (as shown in Figure 2-5) of ECG concentration distribution pcolors are extracted, from figure As can be seen that the first leaf position blade and blade has significant area with the gray scale of other leaves position blade in R component figure and B component figure Not.

After obtaining suitable model and threshold value, when carrying out actually detected identification, including：

(1) image background is removed so as to extract the target area in image using thresholding method；

(2) gray processing and smoothing processing are carried out to target area；

(3) binaryzation gray level image and carry out Mathematical Morphology Method treatment to eliminate hole；

(4) by the boundary line in canny operator detection images, and polygon approach is carried out, in mark closed sides boundary line The region in portion；

(5) relational model according to characteristic wave bands and concentration predicts the ECG concentration values of each pixel in different zones；

(6) the ECG concentration averages in statistics different zones, then the region be as standard determination using ECG concentration thresholds No is the first leaf position blade.

Finally, tea tree Canop hyperspectrum image is scanned, by spectral reflectivity and the relation of ECG concentration set up, is obtained To tea tree canopy ECG concentration profiles, the method according to above-mentioned steps 1~6 is partitioned into the first leaf position blade.

Claims (7)

1. a kind of tea tree tender leaf recognition methods based on ECG concentration differences, it is characterised in that comprise the following steps：

(1) reflectivity at its characteristic wave bands is obtained for fresh leaves of tea plant；

(2) relational expression based on the catechins ECG concentration in tealeaves Yu characteristic wave bands reflectivity, obtains the fresh leaves of tea plant ECG concentration；

(3) the ECG concentration thresholds based on tea tree tender leaf, tender leaf identification is carried out to the fresh leaves of tea plant.

2. the tea tree tender leaf recognition methods based on ECG concentration differences as claimed in claim 1, it is characterised in that the tea tree Fresh leaf from tea tree breed chrysanthemum spring tea tree, the tea tree of Zhejiang agriculture 25 or to meet white tea tree.

3. the tea tree tender leaf recognition methods based on ECG concentration differences as claimed in claim 2, it is characterised in that characteristic wave bands It is 541nm, five wave bands of 616nm, 654nm, 677nm and 695nm.

4. the tea tree tender leaf recognition methods based on ECG concentration differences as claimed in claim 3, it is characterised in that ECG concentration with The relational expression of wave band reflectivity is：

Y_{ECG concentration}=44.988-0.289 λ₅₄₁+4.205λ₆₁₆-6.82λ₆₅₄+3.632λ₆₇₇-0.77λ₆₉₅

In formula：λ_aRepresent the reflectivity at a nm wave bands, Y_{ECG concentration}Represent ECG concentration.

5. the tea tree tender leaf recognition methods based on ECG concentration differences as claimed in claim 4, it is characterised in that for difference Tea tree, the ECG concentration thresholds are：

Chrysanthemum spring tea tree, ECG concentration values >=79mg/g is the first leaf position blade；

White tea tree is met, ECG concentration values >=64mg/g is the first leaf position blade；

The tea tree of Zhejiang agriculture 25, ECG concentration values >=66mg/g is the first leaf position blade.

6. the tea tree tender leaf recognition methods based on ECG concentration differences as claimed in claim 5, it is characterised in that step (1) institute The blade that fresh leaves of tea plant is the 1-6 leaves position since the terminal bud of tree plant down is stated, centre is free of fish leaf.

7. the tea tree tender leaf recognition methods based on ECG concentration differences as claimed in claim 6, it is characterised in that in step (2) Reflectivity at a nm wave bands is corrected by following black and white plate：

${\mathrm{\&lambda;}}_a=\frac{{\mathrm{\&lambda;}}_0-B}{W-B}\&times;DN---\left(1\right)$

In formula：λ₀It is the sample primary reflection spectrum obtained in hyper-spectral data gathering camera bellows, B is covering camera mirror under total darkness environment The demarcation reflectance spectrum of head collection, W is white standard reflection spectrum, and DN is multiplication factor.