FI104128B - A procedure for identifying organisms containing chlorophyll - Google Patents

Description

This invention relates to a method for identifying chlorophyll-containing organisms according to claim 1.

Combined with the mapping of all plant individuals growing in the field, which can be done with the same device as plant identification, the method allows a map of the field to be unique to the species. By means of such a map 10, for example, herbicides can be targeted to the desired plant species.

Herbicides are a major threat to the environment. Agricultural research also aims to reduce the use of herbicides because of the high cost of 15 herbicides. In conventional herbicide application methods, much of the herbicide is applied to the soil instead of the plants to be destroyed. Thus, methods have been developed to target the herbicide to weeds. Early methods were generally based on mechanical detection of weeds higher than the crop. For example, a device is known from U.S. Pat. No. 3,959,924 in which a rod is hung in front of the herbicide spray which triggers the herbicide spray when it touches a high weed. Equipment for weed control and thinning in row cultivation, where '25 herbicide spreaders are controlled in a row based on mechanical or optical observation of plants, are described in U.S. Patent Nos. 3,609,913 and US 1,193,963.

* 30

In recent years, systems have also been developed to distinguish a plant cover from the ground based on the * reflection spectrum characteristic of the plant cover, or essentially the green color of the leaves' (Stafford and Miller 1993, Computers and Elect. In '15: Agric. 9, 217-229). The information thus collected is then used: either directly to direct the herbicidal spray or from the data. collecting a database, utilizing a satellite positioning system, f • «· 2 104128 to distribute herbicides into the field. Plant reflectance spectrum devices are poor in resolution because of the small influence of small plants on the reflectance spectrum, and because the ground surface also significantly reflects the 5 wavelengths at which the reflectance spectrum of the soil and plant cover are the largest (AW Hooper, GO Harries, and B. Ambler 1976). Eng. Res. 21, 145-155; WL Felton 1995, Proc. Brighton Crop Prot. Conf. -Weeds). The resolution of the reflection spectrum method also depends, e.g.

10 ground quality and weather conditions; for example, morning dew significantly reduces the difference in ground and plant cover reflectance spectra at key wavelengths (Felton, m.t.), and it is virtually impossible to distinguish some plants from some soil types (Hooper et al., m.t.).

15

Plant location based on chlorophyll fluorescence has been suggested in the past, though not for herbicide application purposes. For example, from patent application FI914249, a method is known in which seedless lumps in honeycomb assemblies containing 20 seedlings are identified as having no chlorophyll fluorescence. ; by. The use of chlorophyll fluorescence to measure plant stress states and other physiological parameters is also known, for example, from G.H. Krausen and E. Weis (1991) •• 25 articles (Annu. Rev. Plant Physiol. Plant Mol. Biol. 42, 313-349).

»: • · •» »: Minimizing herbicide application rates and maximizing efficacy and more broadly the so-called. It is important for the purposes of precision farming (see S. Blackmore 1994, 30 Outlook on Agric. 23, 275-280) to develop methods capable of mapping and sorting:. identify the plants growing in the field. Species recognition is not • · · * ... possible by reflection spectrum methods,. because the species-specific differences in the reflection spectrum are small 3.5 ·· (Hooker et al., m.t.) and because the size of the plant affects: ': a decisive combination of ground and plant reflection spectrum. Species identification has been attempted with the aid of image analyzers but this method, which requires highly efficient data processing equipment, has not worked in practice (Stafford and Miller, m.t.). Chlorophyll fluorescence spectrum has previously been shown to have inter-species differences that could potentially be used to identify species (E.W. Chappelle, F.M. Wood, Jr., J.E.McMurtrey III and W.W. Newcomb 1984, Applied Optics 23: 134-138). However, measuring the fluorescence spectrum requires a highly sensitive fluorescence detector, and the differences in the fluorescence spectrum 10 have not been shown to be able to identify plants from more than two species.

It is an object of the present invention to provide a map of a plant specimen, in which the plant specimens are classified on the basis of 15 phylogenetic groups, for example species. The method for identifying plants according to claim 1 readily distinguishes between plant species.

In a typical embodiment of the invention, the plants are first located by taking advantage of the fact that other natural objects emit little or no visible light while the plants contain. . chlorophyll a fluoresol. Thus, the fluorescence of chlorophyll a indicates that there is a plant in a particular location. Each localized plant is identified by subjecting it to a series of light treatments, monitoring the variations in fluorescence of .25 chlorophyll a over time during light treatments, and comparing the resulting curve • · | '·· to the corresponding curves measured from known species.

• f · ♦ · ·

In addition to plants, the method can also be used to identify other chlorophyll-containing organisms.

More specifically, the method of the invention is characterized in what is stated in the characterizing part of claim 1 * · '.

· 35:

The invention provides considerable advantages.

4, 104128

The invention enables the automatic identification of plant species. In this case, the plant handling devices can be adjusted to handle different plant species in different ways. For example, a herbicide spray can be constructed to treat each herb species with the 5 herbicides that will be most effective against that species and leave the crops untreated. The invention can also be used to construct the equipment required for agricultural and ecological research, since the preparation of a species map identifying plant individuals is greatly facilitated by the invention. Instead of a species assay, the invention can be used to classify the target organisms into phylogenetic groups that are broader than the species, for example, crude seeds and bare seeds, or smaller species such as varieties. The invention can also be used to identify contacting microorganisms such as cyanobacteria, thereby facilitating, for example, monitoring of the status of water bodies by means of sample organisms.

The invention will now be described in more detail with reference to the accompanying drawings.

20

Figure 1 shows the top surface of a pine needle bundle and a birch leaf. an embodiment of the invention used in testing the method for comparing relative fluorescence yield variations. during the light treatment.

The method of identification of chlorophyll-containing organisms of the present invention is based on the measurement of changes in chlorophyll a fluorescence emission: Y: quantum yield. The term 'relative fluorescence yield1' is used herein to refer to the measured intensity of fluorescence excited by a constant excitation light; in embodiments of the invention, changes in intensity so measured that are proportional to the fluorescence quantum yields are used. changes. Chlorophyll a is present in higher plants, 35 '·; sagebrush, moss, most algae, and cyanobacteria. Almost all chlorophyll in higher plants !! The emission of the Y-containing moieties is the fluorescence of chlorophyll a; Algae and cyanobacteria also exhibit emission of other pigments. Govindjee (1995) has summarized the fluorescence of chlorophyll a (Aust. J. Plant Physiol. 22: 131-160).

The absolute quantum fluorescence of chlorophyll a is relatively small in the plant, at most about 5% of the absorbed light. In a living plant, the fluorescence of chlorophyll a is suppressed by e.g. photosynthetic reactions of photosynthesis that compete with fluorescence for 10 excitation energies absorbed by chlorophyll. At the Earth's surface temperatures, almost all of the fluorescence of chlorophyll a is emitted from the photosynthetic photosynthesis II (PSII) chlorophyll molecules. PSII is functionally the 'first' of the two photosynthetic chain of photosynthesis, since this photoreaction feeds its electrons 15 from the water into the electron transfer chain between the photoreactions, and the reductant produced by the photoreaction I at one end of the chain acts as a carbon dioxide reducing coupler.

At maximum power, the photochemical reactions of PSII attenuate fluorescence to the quantum salt to about one fifth: the maximum value that can be measured by completely stopping the photochemical reaction, for example, by treating the plant with a diuronic herbicide. When the plant is kept in the dark for a while and then suddenly illuminated, the photochemical reactions of PSII begin. However, PSII is capable of delivering electron transfer jets to the • • • '· electrons of the june faster than the chain is initially able to pull, causing the photochemical reaction of PSII after the dark period to abruptly slow down. The momentary slowdown in the photochemical reaction 30 is reflected in the fluorescence rise peak.

Competition between the photochemical reaction and the fluorescence excitation; ·. energy makes fluorescence an important method of research in plant physiology (Krause and Weis, m.t.).

«·«

«I

35 * 1 The fluorescence rise peak described above has a characteristic formula, the fluorescence induction curve, or so-called. Kautsky curve. The formula for the Kautsky-V curve is the same for all plants. The various parts of the Kautsky curve ♦ 6 104128 are influenced by many features of the photosynthetic machine. For example, the steeper the rising portion of the curve, the higher the light-collecting so-called. the antigen pigment system of the test plant PSII has. In fact, the curve reflects the sum of so many features of the photosynthetic machine that, despite years of intensive research, parts of the curve are quite poorly understood. So far, botany has focused on studying features of the fluorescence induction curve that are common to all plant species. For example, O. Björkman and B. 10 Demmig showed (Planta 170, 489-504, 1987) that the so-called.

the ratio of variable to maximum fluorescence is the same for all higher plants. The identification of the plant species subject to the present invention is based on the fact that the plants exhibit species-specific differences in physiological properties. Some of these 15 features can be observed as differences in the Kautsky curve, and differences can be amplified by using a series of different exposure periods to measure fluorescence induction.

The method for identifying organisms containing chlorophyll is as follows.

,,; The identified plant or other chlorophyll-containing organism is shaded by bright natural light for at least the time of identification. The subject is illuminated by an excitation light having a wavelength peak below 700 nm, preferably about 450 nm. The fluorescence of chlorophyll a excited by the excitation light '' is measured by a light detector, directed at • ·: '··, which can be a photodiode or • 9 photomultiplier tube. The detector is shielded from the wavelength band of the excitation light as a filter, which only emits light that is longer than the wavelength of the excitation light 30 and falls on the chlorophyll fluorescence wavelength band. Start a series of one or:. cycles of multiple light treatments that cause Kautsky curve • i /: · type changes in the subject's relative fluorescence yield.

• · '. . Changes in fluorescence yield are directly proportional to · · 35 ·! with changes in fluorescence intensity measured by the detector.

: The information produced by the detector is digitized and processed by some pattern recognition method. For species identification, a series of corresponding signals from any species present in the field has been pre-generated, and pattern recognition occurs by comparing the newly received unknown signal with previously known ones. For the use of pattern recognition methods 5 for analyzing and diagnosing biosignals, see p. A. Cohen Biomedical Signal Processing. Butter. II. Compression and Automatic Recognition (CRC, Boca Raton, 1988).

The excitation light is preferably flashing, i.e., amplitude-10 modulated light, whereby the direct effect of the light treatments on the fluorescence intensity can be filtered out and only their effect on the relative fluorescence yield can be observed. The use of amplitude modulated excitation light to measure the fluorescence of chlorophyll a is known, for example, from DE 3518527 A1.

Typical uses of the identification method of the present invention require that the identification objects be located prior to the beginning of the identification process. Detection is performed on the same or 20 similar types of equipment as detection by mapping the presence of chlorophyll fluorescence in the area under study.

; Also in positioning the fluorescence excitation light 4 'is preferably amplitude modulated light, so that for example' '' sunshine does not interfere with the mapping. The emission of chlorophyll a • 25 in the wavelength of the fluorescence indicates that there is a chlorophyll-containing organism in the field of view of the light detector.

• ·: '·· Information on the position of an organism is stored, for example, using the • 1! .1 · satellite positioning system in a location database.

Chlorophyll fluorescence is the method used to locate plants for two reasons. First, the same device can both detect and recognize the plant in general. Second, plants «· ♦ can be more sensitive to fluorescence than, for example, j · ·

Reflection spectrum described by Stafford and Miller (m.t.) «3.5 ·. based methods. Higher sensitivity is based on the absence of significant wavelengths of chlorophyll a (680-750 nm) in natural objects other than plants •

I 4 M.

• · I

I · · · 8 104128 fluorescence emission that could be excited by light absorbed by chlorophyll. Therefore, virtually all chlorophyll a fluorescence in the terrain is a sign of the plant. The fluorescence intensity is directly proportional to the intensity of the excitation light, so that for plant locus j, the method can be adjusted to almost arbitrary sensitivity simply by using excitation light that is so bright that plants of the desired size are detected.

10 Sunlight can also be used as an excitation light to locate plants. In this case, the wavelengths (e.g., above 670 nm) that i are used to measure the chlorophyll fluorescence must be filtered off from the sunlight spectrum. Because the intensity of sunlight varies, it is necessary to monitor the intensity of sunlight simultaneously with the measurement of fluorescence. This is best done by pointing two light detectors at the same point in the area to be examined, one sensitive to the wavelengths of the fluorescence and the other to the wavelengths transmitted by the excitation light filter.

20 The excitation light compensation system based on the measurement of reflected light works well when using artificial light to excite the fluorescence of chlorophyll a (E. Tyystjärvi and J.

I /, · Karunen 1990, Photosynth. Res. 26: 127-132).

I '

«I

'; 2p The data generated by the plant mapping can be immediately transferred; · »as actuator control information. For example, the herbicide applicator • · ί 'may be commanded to open the nozzle if there is • · V under the detector plant, or to close the nozzle if the plant is not detected.

The information may also be stored, for example, by means of a GPS-30 satellite positioning device in a database for controlling the actuator.

M »« • *, ·; · The technical performance of the detection method was tested under laboratory conditions for 4 hours. as follows.

(M

The testing was carried out in collaboration with the inventor (docent Esa Tyystjärvi, «i

Department of Biology, University of Turku) and assistant Antti Koski

fM · f. I

"f * Γ t« I · 9 104128 and Professor Olli Nevalainen (Department of Computer Science, University of Turku). Dandelion and birch leaves, spruce and pine needles were collected from the terrain. In addition, 5 pumpkins, peas and rye were grown in controlled chambers. dark period between the wet paper, followed by 200 for each type of fluorescence was measured induktiokäyrää. induktiokäyrät rye and pumpkin leaf pieces were measured, fir, pine needle curves 10-20 of stacks 10, and curves of other types of individual newspapers. birch, pumpkin seeds and the curves measured from two side leaves , because the curves of the leaf are different, only the top surface curve of the dandelion was measured, since the dandelion is always upside down in the terrain.

A fluorescence meter was equipped with an amplitude modulated excitation light instrument for measuring the relative fluorescence yield of chlorophyll a (PAM-101, manufactured by Heinz Walz GmbH, Effeltrich, Germany). The fluorometer was controlled by 20 FIP fluorescence programs produced by QA-Data Oy (Turku) and an ADC-12BN4 computer card, which also digitized the analog signal of the fluorometer. Each measurement began with the meter placing a leaf or piece of leaf in a dim room on a fluorometer sensor that illuminated the sample; with dim amplitude modulated light. A piece of leaf caused the 25th relative fluorescence to rise on the fluorometer signal. above the noise level, whereupon the control program initiated a series of light treatments.

The '' '' series of light treatments was as follows. The numbers in Figure 1 refer to the starting time of each of the 30 steps: first a step of 0.8 seconds, when the fluorescence caused by the excitation light flickering at a frequency of only 1.6 kilohertz was measured (1). The second step was to raise the flicker frequency of the excitation light to 100 kilohertz (2). A half second after the start of the second phase of 3.5 lit red LED light (wavelength: 660 nm; photon flux density of 60 omol m'2s-1) in (3) and five seconds' ·]; this was ignited for a further two seconds as bright white "I * · 10 104128 light halogen lamps (photon flux density 5500 omol m "2s'1) that the relative fluorescence yield of all species reached the maximum level (4).

Changes in relative fluorescence yields from light treatments were normalized so that the maximum fluorescence value in each signal was one. Eight different regression lines were fitted to the targets for each curve, and the sixteen parameters of these lines were used as input values for 10 pattern recognition algorithms. For each set of measurements, 100 curve parameters were used as training material for the pattern recognition algorithm and the method was tested with 100 curves. Table 1 shows the recognition results obtained when using pattern recognition R.J. The neural network algorithm described by Schalkoff in Pattern Recognition: 15 Statistical, Structural and Neural Approaches (John Wiley & Sons, Singapore, 1992).

«· * · ♦ • · · •« · »« · »I> III · I · 11 104128

Table 1. Laboratory-specific detection accuracy test. Plant species were identified by the method use method described above. The adaxial (y) and abaxial (a) surfaces of birch, pumpkin and pea leaves were classified as their own. Correctly identified cases account for 96.6%.

Species Identified as a species

Birch Pumpkin Six Pea Pine Rye Dandelion 10 (y) (i) 7y) (S) ly) (5)

Birch (y) 96 3 0 0 0 0 0 0 1 0

Birch (s) O 97 0 1 0 0 O 0 0 3

Pumpkin (y) 1 O 98 0 0 0 1 0 0 0

Pumpkin (a) 0 1 0 97 0 0 0 0 1 1 15 Six 0 100 98 0100 0

Herne (y) 0 0 0 0 0 93 7 0 0 0

Pea (a) 0 0 0 3 0 1 95 0 1 0 Pine 0 000 100 99 0 0

Rye 0 000 1020 97 0 20 Dandelion 0 400 00000 96 - / -: In field conditions the invention is used, for example, as follows.

25 The detection method is used with plant mapping equipment, which includes an illumination and detector unit, a shade, and GPS • satellite positioning equipment mounted on a moving platform. The illuminator continuously emits amplitude modulated I · l excitation light and is able to emit continuous light when needed.

The detector unit is connected to measure only the fluorescence excited by the amplitude modulated excitation light. The purpose of the continuous light light source is to provide a series of illumination cycles according to claim 3. There may be several sources of continuous light. The Lavetti will automatically move through the field in advance of 35 planned tracks, for example navigating a GPS system. through. The tracking and tracing device traverses the field twice.

V. In the first round, the occurrence of plants in the field 12104128 is mapped by measuring the regional distribution of the relative fluorescence yield of chlorophyll a. The measurement results are transferred to the location database. In the second round, the plant mapping device moves to the detected plant specimens, 5 stops and recognizes the plant, which information is added to the location database. The location database is used as input information for, for example, a herbicide spray.

Prior to sending the plant mapping apparatus to field rotation 10, a representative sample (approximately 100 samples) of each plant species or other target phylogenetic class present in the area is collected, each of these is treated with light, If a neural network-like pattern recognition algorithm is used, the so-called training is performed. backpropagation algorithm.

The plant mapping device can also move so slowly that both locating and classifying 20 plants is done in the same field rotation. Here, the movement of the platen is stopped when the device detects a relative fluorescence rise above the noise level, and the device is immediately transferred to plant identification by the light treatments according to claim 3.

j 25 I '']; . The pointing device may also be detached from the sensing device, whereby the pointing device i * ··· 'j may be simpler in structure and use sunlight as the fluorescence excitation light.

The plant locating and identification system may also be packaged as part of a herbicidal syringe such that each herbicidal nozzle is matched by a fluorescence detector having a field of view of the same size j: '·· as the nascent radius of the herbicidal jet. In this case, the equipment must also include a sensor which: 35: indicates the speed of movement of the equipment. The plant locator and I * t '. ·' Detection unit, the speed sensor and the herbicide spray nozzles are connected to the control unit. Upon detection of the poison by the detector, the control unit of the plant to be poisoned will give the corresponding nozzle a timely opening command.

The foregoing has focused on identifying plants by means of the invention.

5 The fluorescence of chlorophyll a can be used to locate and classify not only taller plants, but also mites, mosses, algae and cyanobacteria. The identification of plants or micro-organisms by the method may also be carried out under laboratory conditions. The method 10 of the invention can then be used, for example, to identify naturally occurring microorganisms by means of laboratory samples.

• • • • * • • • • • • •? • · · • I · • · «· •« «« «f

Claims (6)

14 104128 1. A method for automatically identifying a cultivated or herbivory individual to a particular species or to a broader species of phylogenetic group, characterized in that the identification is by subjecting the plant to a series of one or more light periods, measuring the relative fluorescence yield of chlorophyll a emitted from the plant. as a function of illumination and comparisons thus made; fluorescence yield curve from representative phylogenetic groups to similarly measured curves using a pattern recognition algorithm. Identification method according to claim 1, characterized by | that the variation in the relative fluorescence yield of chlorophyll a is measured using amplitude modulated excitation light. . . Plant mapping method based on the identification method according to claim 1, characterized in that the plants to be identified are located by the fluorescence of chlorophyll a prior to their identification. A plant mapping method according to claims 1 and 3, characterized in that the positioning process uses sunlight as the fluorescence excitation light, which is corrected for by the effect of 30 ° intensity variations simultaneously with · · · fluorescence and dividing the measured fluorescence · · • * '· * intensity in sunlight intensity. M) M 35 .:. Method according to claims 1 and 3, characterized in that the information collected by the method is used to control a plant processing device. 15 104128 Use of the method according to claims 1 and 5, characterized in that information on the organisms to be located and identified is collected in a database by means of a satellite positioning system. 5 • • • • • • • • • • • •••••••••••• 104128