"NON-DESTRUCTIVE METHOD AND DEVICE TO CALCULATE VIGOR AND VEGETATIVE EXPRESSION IN TREE-SHRUB VEGETATION AND

APPLICATION THEREOF"

Technical domain of the Invention

The field of the present method is the Wine-growing sector, aiding the vegetative/productive balance assessment of grapevines, by means of the calculation of vigor (defined in wine-growing scientific area as "vigor estimate") and of vegetative expression. The vigor determination is one of several parameters determined in technical scientific investigation, which reveal the vegetative/productive balance in grapevines, being directly related with the life span thereof and productive and qualitative capability of the grapes.

State of the Art

Grapevine vigor is associated with the dynamics of its growth, being characterized by large-diameter and large- length canes, having long internodes, being also associated with a high granddaughter production, green shoot and side shoot production. The leaves also present a larger surface and dark-green shade (Champagnol, 1984; Fregoni, 1999; Magalhaes, 2008, Toda, 2008) .

The vigor estimate is one of several wine-growing parameters, which indicate vegetative/productive balance in a grapevine, and consequently the quality thereof (Toda, 2008) . A reasonable pruning in terms of load and corresponding vigor, thus guarantees not only the best ratio between vegetative and productive parts for a good maturation aiming for quality standards, but also the potential for regular production throughout the years, as well as a vegetative state corresponding to a larger life span of the grapevine (Magalhaes, 2008) .

A reasonable pruning, aiming for a medium vigor, should set up the maximum admissible load compatible with a minimum reduction of the grapevine's capability, Toda (2008) referring that the number of buds distributed per trunk directly influence the foliar area, yield, quality and consequently the microclimate of vegetation cover.

A higher productivity is observed, the lower the vigor will become, in other words, the foliar area is smaller per green shot as well as the ratio thereof with production volume ( Champagnol , 1984; Toda, 2008) . The amount of carbohydrates synthesized on such year is channeled at a larger proportion for a superior number of bunches, thus having a negative effect on the amount stored in reserves and, consequently, on the grapevine's production capability in the following year. Magalhaes (2008) further discloses that a low productivity associated with insufficient load, does not necessarily result in improved vintage maturation and quality, due to the small foliar area and high vigor per cane .

Whenever the vigor is very high, given great fertility conditions in the soil or severe pruning, the excessive densification of the subsequent vegetation (Murisier & Zufferey, 2005), creates unfavorable microclimatic conditions to the differentiation of base buds (low temperature and light intensity) (Reynolds et al, 2007) . On the other hand, whenever the vigor is very low, given a low fertility of the soil, excessively long pruning or diverse illness/accidents, the differentiation is equally affected, however, due to the lack of carbohydrates (Colova et al, 2007), which are produced by the scarce foliar area and the accumulated reserves in the lively parts of the trunk (Kliewer & Dokoozlian, 2005) .

One might then say that, depending on the vigor and vegetative expression observed on a grapevine, one should set up the ideal load, so that on the one hand the maximum productive capacity of the grapevine is exploited and, on the other hand, its qualitative potential is not exceeded (Leeuwen et al, 2009) . This is obtained by means of a medium vigor.

The vigor estimate (expressed in wine sector by the medium weight of a pruning cane) and the vegetative expression (expressed in wine sector by weight of pruning timber), are therefore indicative of the entire guidance system state (Oliveira, 2003) .

According to Figure 2, a grapevine consists of a trunk (7), arms (8) and canes (6). The canes consist of base buds (9), which guarantee, year after year, a new bud burst of the grapevine and consequently new canes.

Annually, during the grapevine's dormancy (period in which the grapevine is exempt from leaves), it is essential to undertake pruning procedures. This operation, consists of removing (cutting) the canes that grew on that year, with the resource of a pruning scissors, maintaining the trunk, arms and only the portion of the canes with acceptable number of base buds for the development of new canes (from the base buds), which are considered to be the ideal to guarantee a good grape production and quality. The number of base buds remaining in each grapevine corresponds to that of a medium vigor, for this is the only alternative for a good life span thereof, as well for a good grape productivity and quality.

In scientific terms, when assessing the vigor of a grapevine, we take action as follows:

1) by the time of pruning operation, which is carried out on the field and with the aid of pruning scissors, a scale and data registration form, the grapevine is pruned (removing of the canes that grew on that year, leaving the grapevine mainly with the cane portion having the number of base buds which are considered as developing towards a balanced vigor in the following year's bud burst) .

2) the canes are removed from the wires and are subject to counting and weighing with the aid of a scale. The values are registered on paper.

3) in office, the "vigor estimate" is calculated and expressed in grams, in other words, the medium weight of a cane is calculated (total weight of pruning canes, also designated vegetative expression, divided by the number of canes) .

This procedure, which is used in the technical scientific area of the wine-growing sector, besides being a "vigor estimate", insofar as there is always a portion of the canes which is not removed (portion with base buds remaining in the grapevine to guarantee the following year's bud burst), is a method which:

1) is time-consuming given that, at a scientific level, field tests often require the calculation of the "vigor estimate" in at least 50 grapevines.

2) depends on the pruner and on the pruning date of the grapevines, in other words, the grapevine vigor must be frequently determined, but as the pruner is not available to undertake pruning procedures on the desired date, the team must therefore wait for the availability to take pace.

The calculation of the "vigor estimate", based on the medium weight of a pruning cane, is directly related to the diameter of the canes, that is:

By the time of the pruning, one might find 3 situations:

1) the canes have large diameter (on-field visual observation) , and the vigor is empirically estimated as being at a high level, and when pruning the grapevine, the number of buds is increased when compared to that which was left in the previous year, so that the grapevine presents a medium vigor in the following year.

2) the canes have medium diameter (on-field visual observation) , and the vigor is empirically estimated as being balanced, and when pruning the grapevine, the number of buds left is equal to that of the previous year, so that the grapevine maintains a medium vigor in the following year.

3) the canes have reduced diameter (on-field visual observation) , and the vigor is empirically estimated as being at a low level, and when pruning the grapevine, the number of buds is decreased when compared to that which was left in the previous year, so that the grapevine presents a medium vigor in the following year.

Depending on the existing methodologies, the expressed value in grams of the "vigor estimate" for the three situations hereinabove described is only made possible subsequently to pruning operation, in other words, after the number of buds left in the grapevine having been determined, by visual observation.

WO9917606 discloses a method and device allowing the determination of the time when a vegetation treatment should take place, by applying fertilizers in the soil and pulverizing the plants with pesticides, depending on the state of the vegetation. The method used in the said application is based on the electromagnetic radiation level on the image pixel which, subsequently to obtaining an image by means of a CCD camera (charge-coupled device) and processing thereof, allows dividing the image in spectral subareas and, based on the amount of vegetation per subarea, the treatment and maintenance of the vegetation is determined .

WO9612401 discloses a plant pulverization controller, based on a probe which, by means of a RGB-probe-based method (RGB - Red, Green, Blue) , allows determining the moment when a plant is to be pulverized. The object is to determine whether the area exceeds a certain color level, thus allowing a pixel-comparison-based assessment on whether the plant is green and whether it requires pulverization. CN101298986 discloses a method allowing the measurement of broad-leaved plant area and allowing the measurement of plant dehydration, based on a digital image of a leaf area. The main application of this method is suitable for scientific investigation.

UA21864 discloses a method allowing the determination of the foliar area in a plant. The method is based on digital images of the plant leaves and on the image of a reference sample, foliar area being calculated upon the number of pixels within the images.

Description

The present invention refers to a novel method for calculating the vigor and vegetative expression in a grapevine, based on a digital image, obtained prior to pruning. The image of the grapevine is obtained on the field, using a digital camera, there being no need for the camera to have special features. The image is subject to digital processing so as to allow identifying, locating and isolating a reference mark, with known dimensions which will give way to determine the area of a pixel within the image and the grapevine canes, which will in turn allow the calculation of the total area of the canes. Afterwards, the medium area per cane is determined and the vigor is calculated, based on a medium area per cane and medium weight per cane ratio. This ratio was obtained from tests comprising the measurements of grapevine cane areas and the weights thereof. The vegetative expression is calculated by multiplying the obtained vigor value by the number of grapevine canes. The present invention aims at providing a novel non ^¬ destructive method, based on digital images, which allows determining the vigor and vegetative expression, simultaneously reducing the time-consuming work associated with on-field data registration, further aiming at exempting from:

1) the pruner .

2) time restriction (day).

3) equipment (scale, pruning scissors and registration paper) and workers aiding in on-field cane weight determination .

With this novel method, one skilled in wine-growing sector may on his own, with resource to a digital camera, calculate the vigor and vegetative expression of one (several) grapevine ( s ) , thus avoiding holding time until pruning season.

Detailed description of the invention

The method of the present invention consists of calculating the medium weight per cane (vigor) and the total weight of the canes (vegetative expression) in a grapevine from a digital image.

The different phases of the method for the determination of the medium weight per grapevine cane (vigor) and total weight of the canes (vegetative expression) are hereinafter presented :

1. Obtaining the digital image

The images are acquired on the field (vineyard) . For obtaining thereof, the grapevines are identified with a number and are prepared in order to retain mainly the yearly canes. For such, all branches stemming from the yearly canes are removed, that is, all granddaughters are removed .

Subsequently, images are obtained using a digital camera, a white cloth, arranged behind the grapevine presenting solely the yearly canes (6), and a 15x20cm reference mark (5) which was placed on a wire, as observed in Figure 3.

In this case, a Sony DSC-H1 was used with a 5.1 megapixel CCD probe. However, since there are no requirements as to the camera's specific features, any camera might be used.

The images obtained on the field have the following characteristics: color images (RGB-Red Blue Green) in JPEG format (.jpg) and 2048x1536 pixels.

2. Image, reference mark and cane processing.

The image processing was carried out using MATLAB Image Processing ToolBox, any other tool having also been possible . a) Concealing of grapevine trunk and arms

In the original image (Figure 3) the grapevine trunk (7) and the arms (8) are painted blue (RGB = 0,0,150), thus a new image being obtained as shown in Figure 4. b) Segmentation of the reference mark

In order to determine the area corresponding to a pixel, one has to isolate the reference mark. For such, a binary procedure is carried out using the red channel (Red) , wherein all pixels with the red channel value superior to 220 correspond to the mark, thus a new image being obtained as shown in Figure 5.

This image is subject to Median Filtering-type filtering with a 10x10 mask for the removal of some eventual irregularity contained therein. The result is a irregularity-free image, as shown in Figure 6.

This image will be the basis for the calculation of the pixel area, which is necessary for the vigor determination (described in 4) .

c) Segmentation of the grapevine canes .

In order to determine the area corresponding to the canes, it is necessary to isolate them from the image (Figure 4) . For such, the following operations are carried out: removal of the reference mark (5), background removal and segmentation of the grapevine canes (6) .

The reference mark (5) is subtracted from the RGB image, in other words, the pixels corresponding to the mark become blank .

Subsequently, the background and the area (10) hiding the trunk and arms are removed, a filtering taking place using the blue channel (Blue) , wherein all pixels having a blue channel value superior to 100 become blank, given that they corresponded to the background. This results in an image as shown in Figure 7.

This image is converted into a grayscale image and is later subject to binary procedures with a 0.5 threshold level. This last operation results in an image, wherein all pixels having a luminance superior to the threshold level become blank and those inferior thereto become black.

The black and white image is applied with a group of masks in order to isolate the canes. Figure 8 shows this group of masks, wherein the marks (represented byH) and respective fillings (represented by ) can be observed.

The application of the said masks results in an image as shown in Figure 9.

From this image, and using the Matlab bwlabel function, all existing polygons in the image are identified and the respective pixel areas are registered. Figure 10 shows the resulting image from the polygon identification, these being represented in different colors.

From this image, the final image showing the canes is created, all polygons whose areas have a value inferior to 1500 pixels being removed. Afterwards, a Median Filtering- type filtering is applied with a 3x3 mask. By means of these operations, a removal of some irregularity that might persist within the image is intended. This results in an image as shown in Figure 11.

3 . Cane medium area per cane and medium weight per cane ratio .

The medium area per cane and medium weight per cane ratio in a grapevine is obtained by means of the following method :

a) Determining the medium weight per cane: this value corresponds to the grapevine's vigor and it was obtained by means of the traditional method. In other words, for each identified grapevine, its respective canes were cut and weighed and the total weight of the canes was divided by the number of canes. cane total

Vigor = Medium Weight per cane =

no. canes

(I)

b) Determining the medium area per cane: for each identified grapevine, all respective canes were scanned. The canes were cut in 20 to 25cm-long portions in order to be scanned in an A4 scanner. The resulting images were processed so as to obtain the total area of the canes per grapevine. The medium area per cane is obtained from the sum of the cane areas in a grapevine divided by the number of canes.

The area of each scanned portion is determined as follows: the scanner area and the matrix size of the scanned image being known, the area corresponding to a pixel within the image is determined;

the scanned image (RGB) is converted into a HSV image and is subject to binary procedure through channel S with a 0.15 threshold level, a black and white image being thus obtained;

the cane portion area corresponds to the number of black pixels multiplied by the area corresponding to a pixel .

The medium area per cane in a grapevine is obtained by the sum of the portion areas of all canes in the grapevine and this value is divided by the total number of canes. portion Area

Medium Area per cane = - no. Canes ^

From the values thus obtained in these tests, a function characterizing the medium area per cane and medium weight per cane ratio was achieved. This function is a second- degree polynomial, as shown in the following equation: f{x) = 0.0047 x ² + 0.2254x + 0.0474

Wherein: X corresponds to the medium area per cane

and

f (x) corresponds to the medium weight per cane. 4 . Determination of the vigor

The determination of the vigor is carried out as follows: a) Determination of one pixel area within the image based on the image resulting from the segmentation of the reference mark (Figure 6) using the following formula:

∑ Black Pixels

Pixel are =

.Mark Area

Wherein: The mark area = 20xl5=300cm ^' b) Determining the total area of the canes based on the final image (Figure 11) using the following formula:

Cane Total Area = (^T Black Pixels )x Pixel area (V) c) Determining the medium area per cane in the grapevine using the following formula:

, , ,. . Cane Total Area

Medium Area per cane = - no. Caries

(VI) d) Finally, the vigor is determined by applying the function found for the Touriga Nacional variety: f {x) = 0.0047x ² + 0.2254* + 0.0474 ( m

Wherein: X corresponds to the medium area per Cane

and

f(x) corresponds to the vigor.

In Figure 12, an example of the results for the vigor determination in a grapevine may be observed.

5. Determining the vegetative expression

The determination of the vegetative expression is calculated by multiplying the vigor value previously obtained by the total number of canes in the grapevine.

Figure 1 shows different blocks, each one having a specific functionality, which is hereinafter described:

1. Obtaining of the digital image (1) : the grapevine image is obtained on the field, using a digital camera;

2. Image processing of reference mark and canes (2) : the image is subject to a digital processing so as to identify, locate and isolate the reference mark, having known dimensions which will allow determining the pixel area within the image and the grapevine canes, which will give way to calculate the total area of the canes;

3. Medium area per cane and medium weight per cane ratio (3) : polynomial ratio previously obtained from the tests consisting of the measurements of the cane areas in the grapevines and weights thereof;

4. Process for vigor calculation (4) : based on the pixel area, on the cane total area, on the number of canes and on the medium area per cane and medium weight per cane ratio, the vigor is thus calculated;

5. Results: vigor and vegetative expression are calculated based on the obtained vigor value and on the total number of canes.

Summary of the invention

The present invention aims at defining a non-destructive method, which is applicable before the pruning, for determining the vigor and vegetative expression in tree- shrub vegetation, which is oriented along surfaces, by means of the following steps:

a) acquiring a digital image obtained on the field;

b) processing the image in order to identify, locate and isolate the reference mark and the canes;

c) determining the area of a pixel within the image;

d) determining the total area of the canes and medium area per cane;

e) applying a medium area per cane and medium weight per cane polynomial ratio; and

f) determining the vigor and vegetative expression, based on the pixel area, on the total area of the canes, on the number of canes and on the ratio obtained in step e) . In a preferred embodiment of the present method, the processing of the image carried out in the previously- mentioned step b) comprises the following steps:

i. concealing of the trunk and arms;

ii. segmenting the reference mark for binary procedure; iii. segmenting the canes.

In another preferred embodiment, the medium area per cane and medium weight per cane polynomial ratio, referred to in e) of the disclosed method, is obtained by the following steps :

i. determining the medium weight per cane;

ii. determining the medium area per cane;

iii. determining the polynomial.

Still in another preferred embodiment, the medium area per cane and medium weight per cane polynomial ratio, referred to in e) is as follows:

f {x) = 0,0047 x ² + 0,2254 + 0,0474

wherein :

x corresponds to the medium area per cane and

f (x) corresponds to the medium weight per cane.

In a preferred embodiment, the polynomial ratio is applied to the Touriga Nacional variety.

Another objective of the present invention is a device for implementing the previously-defined method, comprising the following elements:

- digital photograph camera;

- processor unit;

- display; and - operating system.

It is still an objective of the present invention to use the method and device defined in vegetative/productive balance assessment in tree-shrub vegetation; in tree-shrub life span assessment; and in productive and qualitative capability assessment in tree-shrub vegetation.

Examples

For an easier understanding of the invention, examples of preferred embodiments of the invention shall be hereinafter described, which do not however intend to limit the scope of the present invention.

Example of the polynomial ratio determination using 6 Touriga Nacional grapevines

In order to obtain the medium area per cane and medium weight per cane polynomial ratio in a grapevine, tests on 6 Touriga Nacional grapevines in Vineyard Site Douro were carried out.

The selection criterion of the grapevines was based on the condition that these should have a low, medium and high vigor estimate in order to embrace the whole range of values in the test.

The medium area per cane and medium weight per cane polynomial ratio in a Touriga Nacional grapevine was obtained by means of the following method:

b) Determining the medium weight per cane: this value corresponds to the grapevine's vigor, being obtained by means of the traditional method. In other words, for each identified grapevine, its respective canes were cut and weighed and the total weight of the canes was divided by the number of canes.

,„■ , cane total weight

Vigor = Medium Weight per cane =

no. canes

The results thus obtained are listed in the following table :

Table 1 - Vigor estimate by traditional method

Vine No. canes Total Weight Weight/no. canes

(gr . ) (gr . )

Vine .2 12 266.81 22.23

Vine .3 10 639.76 63.98

Vine .5 13 959.68 73.82

Vine .4 9 811.26 90.14

Vine .6 9 1042.01 115.78

Vine .1 8 1103.89 137.99 c) Determining the medium area per cane: for each identified grapevine, all respective canes were scanned. The canes were cut in 20 to 25cm-long portions in order to be scanned in an A4 scanner. The resulting images were processed so as to obtain the total area of the canes per grapevine. The medium area per cane is obtained from the sum of the cane areas in a grapevine divided by the number of canes.

The area of each scanned portion is determined as follows: the scanner area and the matrix size of the scanned image being known, the area corresponding to a pixel within the image is determined;

the scanned image (RGB) is converted into a HSV image and is subject to binary procedure through channel S with a 0 . 15 threshold level, a black and white image being thus obtained;

the cane portion area corresponds to the number of black pixels multiplied by the area corresponding to a pixel .

The medium area per cane in a grapevine is obtained by the sum of the portion areas of all canes in the grapevine and this value is divided by the total number of canes. portion Area

Medium Area per cane =—

no. Canes

The results thus obtained are listed in the following table :

Table 2 - Medium area per cane for each grapevine

Vine No . canes Total Area Area/ no . canes

(cm ² ) (cm ² )

Vine .2 12 580 . 22 48 . 35

Vine .3 10 952 . 68 95 . 2 7

Vine .5 13 13 79 . 23 106 . 09

Vine .4 9 1029 . 77 114 . 42

Vine .6 9 1202 . 26 133 . 58

Vine .1 8 119 7 . 40 149 . 68 d) From the values obtained in these tests, it was possible to conclude that the behavior of the weight, dependent on the number of canes, was identical to that of the area depending on the number of canes, as shown in Figure 13 . e) Given the result described in the previous paragraph, it was made possible to conclude that there should have been a correlation between the medium area per cane and the medium weight per cane. Therefore, a function corresponding to a second-degree polynomial was found, as shown in Figure 14.

In other words, the polynomial:

f {x) = 0,0047 x ² + 0,2254 + 0,0474

represents the medium area per cane and medium weight per cane ratio (vigor), wherein: x corresponds to the medium area per cane and

f (x) corresponds to the medium weight per cane.

Table 3 lists the values, for each grapevine, of the medium weight per cane calculated using the traditional method and values obtained using the polynomial f (x) .

Table 3 - Medium weigh per cane using the traditional method and polynomial f (x)

Vine Area/no. canes f(x) = weight/no.canes Weight/no. canes

(cm ² ) (gr . ) (gr . )

Vine .2 48.35 21.93 22.23

Vine .3 95.27 64.18 63.98

Vine .5 106.09 76.86 73.82

Vine .4 114.42 87.37 90.14

Vine .6 133.58 114.03 115.78

Vine .1 149.68 139.08 137.99

The obtained values using the traditional method and the polynomial present a 0,998 correlation. Practical example of method application in the determination of vigor and vegetative expression in a Touriga Nacional grapevine

A digital photograph of a Touriga Nacional grapevine was taken on the field and, according to the algorithm shown in Figure 15, this digital image was subject to the following steps :

the trunk and arms of the grapevine (11) were concealed;

the segmented image with the reference mark (12) was obtained;

the value of the pixel area within the image (13) was obtained;

the segmented image with the grapevine canes (14) was obtained;

the medium area value per grapevine cane (15) was obtained;

the vigor is determined ( 17 ) , by means of the polynomial ratio (16) .

The following table shows: the obtained result, vigor determination and the "vigor estimate" result obtained by the traditional method.

Table 4 - Vigor and Vigor Estimate

Vigor Estimate

Vigor (gr.)

(gr.)

24.97 22.23

The vigor estimate is an approximate value because it does not account for the total weight of the canes, but rather to the weight of the canes which have been pruned. The vigor determined using the method, is considered together with the total weight of the canes (vegetative expression) . Therefore, the vigor estimate is always inferior to the real vigor.

Table 5 shows the vegetative expression based on the calculated vigor value.

Table 5 - Vigor, number of canes and vegetative expression

Vegetative Expression

Vigor (gr.) Number of canes

(gr.)

24.97 12 299.64

Description of the Figures

For an easier understanding of the invention, Figures are enclosed, which represent preferred embodiments of the invention and which do not however intend to limit the scope of the present invention.

Figure 1: Schematic representation of the method, wherein:

(1) represents the obtaining of the image;

(2) represents the mark, trunk, arms and canes processing;

(3) represents the medium area per cane and medium weight per cane ratio;

(4) represents the vigor and vegetative expression

determination;

(5) represents the reference mark;

(6) represents the canes.

Figure 2: Representation of a grapevine in bilateral cord orientation, wherein:

(6) represents the canes; (7) represents the trunk;

(8) represents the arms;

(9) represents the base buds.

Figure 3: Representation of the original image, wherein:

(5) represents the reference mark;

(6) represents the canes;

(7) represents the trunk;

(8) represents the arms.

Figure 4: Representation of the image with trunk and arms concealed, wherein:

(5) represents the reference Mark;

(6) represents the canes;

(10) represents the area concealing the trunk and arms.

Figure 5: Representation of the reference mark binary procedure result.

Figure 6: Representation of the reference mark filtering result .

Figure 7: Representation of the image without mark, background, trunk or arms, wherein:

(6) represents the Canes.

Figure 8: Representation of the masks.

Figure 9: Representation of the image after mask

application .

Figure 10: Representation of the image with polygons. Figure 11: Representation of the final image.

Figure 12: Representation of the vigor calculation result.

Figure 13: Representation of weight and area behavior, dependent on the number of grapevine canes.

Figure 14: Representation of real vigor and polynomial.

Figure 15: Representation of the algorithm, wherein:

(11) represents the grapevine's trunk and arms concealment;

(12) represents the obtaining of the segmented image with the reference mark;

(13) represents the obtaining of the pixel area value within the image;

(14) represents the obtaining of the segmented image with the grapevine canes;

(15) represents the obtaining of the medium area per cane value in the grapevine;

(16) represents the use of the polynomial ratio;

(17) represents the obtaining of the vigor and vegetative expression value.