WO2014100856A1

WO2014100856A1 - An image processing based method to estimate crop requirements for nutrient fertiliser

Info

Publication number: WO2014100856A1
Application number: PCT/AU2013/001519
Authority: WO
Inventors: Ahmed K. AL-ANI; Mahdi M. ALI
Original assignee: University Of Technology, Sydney
Priority date: 2012-12-24
Filing date: 2013-12-23
Publication date: 2014-07-03

Abstract

A method and device for analysing a leaf to provide data on leaf content and/or dimensions of said leaf. The device (100) comprises an optical scanner having a power source (1) power/scan button (2) and a scanning aperture (10). The device is adapted to pass over the leaf and capture an image of at least a part of the leaf. A suitable processor then processes the resultant image from the optical scanner (100) to produce data on leaf content and/or leaf dimensions. In one embodiment processor may apply a plurality of algorithms to provide data on leaf dimension, nutrient content and/or chlorophyll status of the leaf.

Description

An image processing based method to estimate crop requirements for nutrient fertiliser Technical Field

[0001] The present invention relates to methods and apparatus for obtaining data concerning plant material and in particular but not only to obtaining estimations of nitrogen and chlorophyll status as well as dimensions and area of plant leaves.

Background Art

[0002] Any discussion of the prior art throughout the specification should in no way be considered as an admission that such prior art is widely known or forms part of common general knowledge in the field.

[0003] There are a number of current conventional systems able to measure various aspects of vegetable matter including leaves. Such measurement of vegetable material such as leafs is extremely important for several reasons. Firstly, it allows a producer to determine the "health" of the leaf including the size of the leaf and the leaf content. In many cases, the nutrients held in the leaf are affected by farming methods such as fertilisation techniques. In addition, it is well known in the art that chlorophyll status of the leaf provides a good measure of the leafs health.

[0004] Currently, leaf dimensions can be determined by many methods. Traditional methods are represented by grid counting and paper weighing. These techniques are simple in principle and can produce relatively accurate results. However, these techniques are time consuming and errors are not uncommon. In addition, some of these manual techniques can cause damage to the leaf.

[0005] Leaf area is an important agronomical parameter as it is related to plant growth and photosynthetic capacity. In predictive research, leaf parameters (area, length, width, average width and perimeter) are very important sources for planting row configurations, trimming and pruning plans and fertilisation programs. As a result, measurements of leaf parameters are very important to study plant biological characteristics and guide agricultural production practice. These measurements are also important for studies of plant nutrition, plant-soil-water relations, plant competition, plant protection measures, respiration, light reflectance and heat transfer in plants. Hence, it is an important factor in understanding photosynthesis, light interception, water and nutrient use, crop growth and yield potential. [0006] Traditional leaf area estimation methods however, are mostly destructive, time- consuming, and costly. This has lead to the development of specialised devices for determining various leaf parameters. One of the most popular devices is the Li-Cor 3100 leaf area meter. This device is quite accurate; however there are some limitations to its usage. For example, large leaves that do not fit in the device have to be divided into smaller portions, and processed individually. However, this process can easily lead to errors.

[0007] Recently, image processing techniques have been developed to measure leaf dimensions using a digital camera with special software like Photoshop or Matlab. However, this approach needs calibrations in order to control light, height and angle variations when capturing photos.

[0008] In addition to leaf dimensions, other aspects of the leaf including its content, nutrient levels, chlorophyll levels etc are important.

[0009] Nitrogen is very important to yield because of its key role in cell division, protein synthesis and enzyme production. If cell division is stopped, leaf area stops increasing and the plant thereby loses its potential to produce a high yield. Despite the importance of Nitrogen supply to plant growth, excessive supply in fertiliser is costly and excess Nitrogen that runs off arable land can have a negative impact on the environment. Thus a mismatch between Nitrogen supply and crop requirement can hamper crop growth and harm the environment, resulting in low Nitrogen -use-efficiency (NUE) and economic losses.

[0010] There are two techniques for foliar analysis of nitrogen content: destructive and nondestructive. It has been shown that plant nitrogen status can be accurately estimated using a destructive technique; in which samples are analysed using laboratory procedures. This technique is generally time consuming, costly and labour intensive. In contrast, nondestructive methods can be rapid and less expensive than the destructive techniques, but are generally less accurate. There are a number of non-destructive methods available that vary in complexity and optimality.

[0011] One of the most widely used techniques is the Konica Minolta Chlorophyll Meter (SPAD-502). This is a hand held device that estimates the chlorophyll content of leaves, as leaf chlorophyll content is closely related to leaf nitrogen concentration. However, SPAD has two main limitations. First, this method is very sensitive to sampling details and errors made at this stage can easily influence the measurement values. This sensitivity is caused by the very small measuring area of the device (around 12.57 mm²), which makes it important to take several readings and average them. Second, since plant chlorophyll is affected by many factors, it is impossible to identify a universally applicable SPAD value that is indicative of adequate foliar nitrogen content for all species under all environmental conditions.

[0012] It is an object of the present invention to overcome or ameliorate at least one of the disadvantages of the prior art, or to provide a useful alternative.

[0013] At least in a preferred embodiment, the present invention seeks to provide a method and apparatus which reliably provides data on foliar dimensions and leaf content in a nondestructive fashion. In a particularly preferred embodiment, the present invention provides a single device which provides both leaf area data and nutrient content data. At least in this embodiment, the present inventive method and apparatus is a significant advance over the prior art and provides a more complete and thorough understanding of plant health in a more consistent and user friendly fashion.

Disclosure of the Invention

[0014] In a first aspect the present invention provides a device for analysing a leaf comprising:

an optical scanner adapted to pass over said leaf and capture an image of said leaf a processor adapted to process said image and hereby provide data on both leaf content and/or leaf dimensions

wherein said device can simultaneously provide both leaf dimension and leaf content data.

[0015] In a second aspect the present invention provides a method of analysing a leaf comprising passing an image capture device over said leaf in near proximity or light contact with said leaf to thereby capture an image of said leaf, and processing said image to provide data on leaf content and/or dimensions of said leaf.

[0016] Preferably, the processor is adapted to processing the image utilising the RGB image content to calculate an expected nitrogen content. In one embodiment, the RGB values are averaged over the leaf image to obtain values RA, GA and BA, and the nitrogen content is derived substantially in accordance with the formula: Nitrogen content = ((RA + GA + BA) / 3 ) - GA.

[0017] In accordance with a further aspect of the present invention, there is provided a method for analysing the leaf content of a plant and particularly nitrogen content, the method including the steps of: scanning an RGB image of the leaf; and utilizing the RGB image to obtain values RA, GA and BA, being averaged values of each of the R, G and B components respectively and deriving the nitrogen content of the leaf derived substantially in accordance with the formula:

Nitrogen content = ((RA + GA + BA) / 3 ) - GA.

[0018] Unlike the prior art, the present invention provides a reliable method and apparatus for providing both leaf dimensional data as well as leaf content. In a preferred embodiment, this data is provided by a simple sweep or scan over a leaf. It is not necessary to take multiple readings or in any way damage the plant to obtain this information.

[0019] In a particularly preferred embodiment, a simple hand-held optical scanner can be used to measure leaf dimensions including size, area, etc, as well as estimate chlorophyll and nitrogen content in plant foliage.

[0020] Further, since the present inventive apparatus and device can use a simple handheld optical scanner held directly adjacent or in light contact with the leaf, it does not suffer from the conventional difficulties associated with digital cameras for instance which must be constantly calibrated to take into consideration current light conditions, angle and height variations when taking photographs. As will be clear to persons skilled in the art, a farmer in the field simply wishes to obtain data on his crop as quickly as possible in a simple and consistent fashion. If a farmer takes a photograph of a particular leaf, the comparison photograph taken at a later time must be done under almost identical conditions, light, height and angles to give any useful comparative data. While there are algorithms which can take this material into consideration, they are both expensive and not always accurate. The present inventive method and apparatus allow an operator in the field a consistent, reliable and automatically calibrated technique for measuring leaf dimensions and leaf content.

[0021] It will be understood that the present invention has application for a wide range of vegetable matters. [0022] It should be understood throughout the specification that the term "foliage", "foliar" and "leaf" refers to any plant or vegetable material.

[0023] As will be discussed below, the present inventive method and apparatus is extremely flexible and can be used for a wide variety of foliar material to determine its dimensions and/or content. As mentioned above, using a simple portable optical scanner an operator can provide significant data directly from the field without the need to provide sample for laboratory analysis. Additionally, the use of such a hand-held scanner avoids the need for multiple specialised devices such as the SPAD-502 Chlorophyll Meter by Konica Minolta and the Li-Cor 3100 leaf area meter

Brief Description of the Drawings

[0024] A preferred embodiment of the invention will now be described, by way of example only, with reference to the accompanying drawings in which:

[0025] Figure 1 A and 1 B are perspective views of a leaf scanner device in accordance with a preferred embodiment of the present invention.

[0026] Figure 2 is a support apparatus in accordance with a further embodiment of the present invention used with leaf scanner device.

Best Mode for Carrying out the invention

[0027] We refer to Figure 1 A and 1 B showing a hand held portable scanner for carrying out the present invention. In this case, the optical scanner used is a PICO portable scanner produced in Australia by Mint Technology. The device is an optical scanner which can be used in colour or monochrome. It has a 600/300 dpi scan selection feature. The scanned image is saved as a J Peg file which is saved to a storage device. In this case the storage device is a micro SD card. The device in question can support micro SD cards up to 32 gigabytes.

[0028] The device 100 suitable for carrying out the present inventive method comprises a power source 1 (in this case batteries) and power/scan button 2 and scanning aperture 10 (on the underside). In the embodiment shown, an indicator 3 shows an error if the scanner for example, is conducted with excessive speed. As will be clear to a person skilled in the art it is desired to pass the device 100 over the leaf in a consistent manner. [0029] In a particularly preferred embodiment the device 100 may be set to high or low resolution. As shown more clearly in Figure 1 B, the device may also include a USB interface 8 for downloading stored data for further processing. Alternatively, a suitable storage device such as micro SD card 9 can be incorporated into the device for later remote downloading.

[0030] Preferably, the device 100 of the present invention is held immediately proximal to and most preferably lightly contacting the leaf to be scanned. In a particularly preferred embodiment, as shown in Figure 2, a light coloured base or support device 200 is provided and the leaf 150 placed on this support base 200

[0031] Preferably the face 20 of the support 200 is provided as a light coloured or white background. Leaf 150 is held against this background by means of a transparent sheet 25. A clamp 30 or similar device can hold the transparent sheet 25 and leaf 150 against the support device 200 if needed. This provides a simple, reliable and most importantly consistent flat surface over which the scanner 100 may be passed to thereby scan a leaf 150 through transparent sheet 25. If desired, marking or graduations can be applied to the face 20 of the support base to assist in alignment of the leaf, however this is not essential.

[0032] Once the portable scanner 100 is activated, it is placed over the support device 200 and passed at the appropriate speed over leaf 150. The resultant scanned images are analysed using image analysis tools. These can be formulated in many different systems, with Matlab chosen by the present inventors as being most suited to prototype implementation. The resultant image is either held in the device for later downloading or in some embodiments can be transmitted to a remote data storage for further processing. Appropriate

analysis/processing of the data image then provides the desired data relating to leaf dimensions such as area etc., or leaf content such as nitrogen, chlorophyll, etc.

[0033] The analysis is basically broken into two parts. Leaf area measurements and analysis of RGB (red green blue) values to achieve maximum correlation with true leaf content.

[0034] As regard estimating the dimensional parameters of the leaf, such as height, width, average width, perimeter and area, a two-dimensional filer is used to separate the scanned leaf from the preferred white background sheet. The leaf area measurements start by preferably applying a digital filter to reduce the effect of outlier pixels. After applying an edge detection algorithm, the leaf area and other dimensions can be calculated. In the preferred embodiment, the leaf is scanned from top to bottom, thereby providing the highest and lowest points of the leaf. This allows suitable calculation of the height. The extreme points to the left and right of the scanned image are used to measure width. The average width, perimeter and area may be determined after identifying the leaf contour.

[0035] The thus determined leaf parameters were compared with a conventional commercially-available leaf meter (Li-Cor 3100 leaf area meter). The results are tabled below in Table 1 below. This clearly shows that the preferred embodiment provides high accuracy as compared to commercially-available products. Indeed, the average error produced according to the present invention is substantially less than the conventional apparatus, (average over 8 samples) This is exemplified in Table 2 below.

Table 1 : Leaf Area Meter

Table 1

Table 2: Leaf length and width measured by proposed method and Li-Cor 3100 compare with actual width and length

Table 2

[0036] As mentioned above, the aforementioned scanning technique can also be used to measure leaf content. As macro nutrients were of most interest, the applicants have presently focussed their efforts on measuring nitrogen levels. However, it will be clear that the present invention has applicability for a wide range of leaf content analyses. In particular the applicant envisages that the present inventive device and method may be used to obtain data on phosphorus content.

[0037] Experiments were conducted under controlled greenhouse conditions on tomato, broccoli and lettuce, with different nitrogen treatments. Seeds of the three species were planted in 75 pots (25 for each species). For each of the three species, the 25 pots were divided into five groups, each group received a pre-specified nitrogen treatment. All pots were filled with vermiculite, and for the first eight weeks they all received the standard level of nitrogen (around 4.3 grams/10 litres) where the nutrient solution was renewed every three days. After week eight, five different nitrogen treatments in the form of NH₄N0₃ were applied for a period of seven weeks. The five groups were subjected to 0%, 25%, 50%, 75% and 100% of the recommended nitrogen treatments.

[0038] Relevant full color images were captured. The RGB values of leaf pixels were then averaged to obtain three values, one for each color component. The following overall formula was then used to obtain an estimated nitrogen or chlorophyll content:

Coptileaf = ((R + G + B) / 3 ) - G

[0039] The first component of the formula was included for normalisation purposes and the second green color component was used due to its relation with the chlorophyll content of plants.

[0040] As a control, the chlorophyll and nitrogen levels were also measured using standard laboratory techniques with nitrogen measured using a TruSpec Leco CH N Analyser and total chlorophyll using spectrophotometry. These results were compared with the present inventive method and the conventional SPAD-502 method. Readings were taken in weeks 10, 13, 15, 18 and 21 for broccoli and 10, 13, 15, 18 for tomato and lettuce . Results are shown in Table 3.

Plant Experiment Liner regression coefficient (R2) Correlation type coefficient (R)

Broccoli New method with N y = 0.5938X + 14.566 / R² = 0.8844 0.9404

lab

Broccoli Ch lab with N lab y = 0.9882X + 4.81 13 / R² = 0.81 15 0.9008

Broccoli SPAD with N lab y = 0.2277X - 4.8281 / R² = 0.7919 0.8899

Broccoli New method with Ch y = 1.3177X - 28.775 / R² = 0.7933 0.890658 lab

0.671233

Broccoli SPAD with Ch lab y = 2.5679X + 19.01 / R² = 0.4506

Tomato New method with N y = 3.1427X - 31 .855 / R² = 0.8742 0.9350

lab

Tomato Ch lab with N lab y = 1.4054X + 1.14092 / R² = 0.9259

0.8572

Tomato SPAD with N lab y = 4.5383x + 1 1.151 / R² = 0.8215 0.9064

0.920722

Tomato New method with Ch y = 2.0734X - 33.791 / R² = 0.8477

lab

0.908585

Tomato SPAD with Ch lab y = 3.1471X + 7.1822 / R² = 0.8255

Lettuce New method with N y = 2.079x - 38.93 / R² = 0.522 0.722808 lab

Lettuce Ch lab with N lab y = 0.809x + 0.889 / R² = 0.545 0.738433

Lettuce SPAD with N lab y = 3.505x + 5.305 / R² = 0.528 0.727138

0.875224

Lettuce New method with Ch y = 2.314x - 40.114 / R² = 0.766

lab

0.738893

Lettuce SPAD with N lab y = 3.3197X + 5.8201 / R² = 0.546

Table 3

[0041 ] As can be seen from Table 3, the present inventive device and method shows a very strong correlation with lab measurements for all three species, namely broccoli, tomato and lettuce. While the SPAD readings achieved high correlation with some species, there are fluctuations. Accordingly, the present inventive technique is at least as accurate as current conventional and well-accepted techniques.

[0042] It can be seen therefore that the present invention provides a significant advance over conventional mechanisms. The inventive apparatus and method provides a simple device capable of measuring various foliar (leaf) dimensions, macro-nutrient levels and chlorophyll.

[0043] It can be seen that the present invention does not require the use of different devices to measure leaf dimensions as well as leaf content in particular nutrient content and even more particularly nitrogen content. Further, leaf dimensions can be measured without the need for excessive calibration, which, in conventional systems, potentially leads to further inaccuracies.

[0044] The use of a conventional hand held colour scanner in accordance with the present invention gives maximum flexibility, ease of use as compared with conventional systems which generally use specifically designed apparatus.

[0045] It will be understood that variations and modifications may be made to the embodiments shown without departing from the spirit or scope of the inventive idea.

Claims

1. A device for analysing a leaf comprising:

an optical scanner adapted to pass over said leaf and capture an image of at least part of said leaf

a processor adapted to process said image and thereby provide data on both leaf content and/or leaf dimensions

2. A device as claimed in claim 1 wherein said optical scanner is a hand held optical scanner.

3. A device as claimed in claim 1 or claim 2 further comprising a base for support of said leaf, said device being adapted to pass over said base proximal to or in light contact with said leaf.

4. A device as claimed in any one of the preceding claims wherein said processor means comprises a plurality of algorithms for application to said captured image to provide data on leaf dimension, nutrient content and/or chlorophyll status of said leaf.

5. A device as claimed in any one of the preceding claims wherein said device is adapted to provide data on nutrient content of said leaf.

6. A device as claimed in any one of the preceding claims wherein said device is adapted to provide data on chlorophyll status of said leaf.

7. A device as claimed in any one of the preceding claims wherein said device is adapted to capturing an image of at least 1 cm² of said leaf

8. A device as claimed in any one of the preceding claims wherein said device is adapted to capturing an image of at least 5cm² of said leaf

9. A device as claimed in any one of the preceding claims wherein said device is adapted to capture an image of the entire leaf area.

10. A device as claimed in any one of the preceding claims wherein said device is adapted to provide data on one or more of leaf height, leaf width, leaf average width, leaf perimeter and/or leaf area.

11. A device as claimed in any one of the preceding claims wherein said device is adapted to provide data on nitrogen content of said leaf.

12. A device as claimed in any one of the preceding claims wherein said processing includes utilising the RGB image content to calculate an expected nitrogen content.

13. A device as claimed in claim 16 wherein the RGB values are averaged over the leaf image to obtain values RA, GA and BA, and the nitrogen content is derived substantially in accordance with the formula:

Nitrogen content = ((RA + GA + BA) / 3 ) - GA

14. A device as claimed in any one of the proceeding claims wherein said device is adapted to provide data on phosphorus content of said leaf.

15. A method of analysing a leaf comprising passing an image capture device over said leaf in near proximity or light contact with said leaf to thereby capture an image of said leaf, and processing said image to simultaneously provide data on leaf content and/or dimensions of said leaf.

16. A method as claimed in claims 14 wherein said image capture device comprises an optical scanner operatively associated with an image processing device.

17. A method as claimed in claim 14 or 15 wherein said processing of said image provides data on nutrient content of said leaf.

18. A method as claimed in claim 14 or 15 wherein said processing of said image provides data on nitrogen content of said leaf.

19. A method as claimed in claim 14 or 15 wherein said processing of said image provides data on phosphorus content of said leaf.

20. A method as claimed in claim 14 or 15 wherein said processing of said image provides data on chlorophyll status of said leaf.

21. A method for analysing the leaf content of a plant, the method including the steps of:

(a) Scanning an RGB image of the leaf; and

(b) Utilizing the RGB image to obtain values RA, GA and BA, being averaged values of each of the R, G and B components respectively and deriving the nitrogen content of the leaf derived substantially in accordance with the formula:

Nitrogen content = ((RA + GA + BA) / 3 ) - GA.