Image processing method, image processing apparatus, and image processing program for seedlings

Abstract

There has been no method for three-dimensionally displaying and diagnosing the internal structure of a minute seedling.　To solve said problem, the present invention comprises: an imaging step S1 for tomographically imaging a seedling which is an imaging subject; a generation step S2 for generating three-dimensional volume data including the internal structure of the imaging subject by sequentially laminating, in a direction orthogonal to tomographic layers, a plurality of captured two-dimensional tomographic image data; an identification step S3 for searching for an equivalent value surface from within the three-dimensional volume data and identifying a curved surface in the three-dimensional shape of the imaging subject; and a conversion step S4 for converting data including the three-dimensional shape into a generically usable format after performing mesh processing.

Description

Seedling image processing method, image processing device, and image processing program

[0001] The present invention relates to an image processing method, an image processing device, and an image processing program for determining the quality of seedlings from images in the agricultural field.

[0002] To improve productivity, farmers purchase seedlings for their own cultivation from seedling producers and retailers. The seedling producers and retailers raise the seedlings to a certain extent before providing and selling them to farmers. Grafting is a common method used by seedling producers and retailers to raise seedlings. Grafting is known to avoid damage from pests and diseases, improve harvest numbers, and stabilize quality.

[0003] Grafting is the process of joining two or more plants together at the cut edges of an artificial cut to create a single plant. The grafted seedling is called the grafted seedling, and the seedling from which the graft is made is called the original seedling. The plant on top of the grafted seedling is called the scion, and the plant on the bottom is called the rootstock. Grafting is often used for vegetables such as tomatoes, eggplants, cucumbers, and watermelons, as well as fruits and flowers. Inside the stem are phloem, which transport nutrients, vessels, which transport water, and vascular bundles, which are made up of phloem and vessels. Grafting is the process of fusing the various tissues, including the vascular bundles, between the scion and the rootstock. The number of days required for fusing is called the curing period.

[0004] Naturally, it is necessary to avoid using seedlings whose stems, specifically their vascular bundles, have been affected by diseases, pests, bacteria, viruses, etc. as source seedlings. Using seedlings whose vascular bundles are free of diseases, pests, etc. as source seedlings is important in order to produce grafted seedlings with a high survival rate (percentage of fusion) and a good yield, that is, high-quality grafted seedlings.

[0005] In addition, the government and the nationwide JA (Agricultural Cooperative Association) notify farmers who purchase seedlings of the planting date. Planting refers to transplanting from one location to another, for example, transplanting grafted seedlings purchased in pots into a field in a greenhouse. One of the reasons for the planting date is to prevent diseases and pests. In order to ensure that farm greenhouses are protected from pests by heat disinfection, the number of days required for disinfection is specified, and the planting date is determined from that. Because the planting date is fixed, orders for grafted seedlings from farmers to seedling producers and retailers are concentrated in a short period of time.

[0006] To meet the intensive demand for grafted seedlings, seedling producers and distributors are trying to shorten the number of days required for seedlings to be cured. To shorten the number of days required for curing, it is necessary to select original seedlings that are expected to have a high survival rate. However, seedling producers and distributors have traditionally only used empirical methods to select and identify such original seedlings. Seedling producers and distributors have been seeking methods, devices, and programs for determining the condition of seedlings (original seedlings) without relying on empirical rules, and such methods are actually known (for example, JP 2008-295407 A and JP 2016-021929 A).

[0007] In JP 2008-295407 A, a single seedling is placed on a pedestal, and an imaging device installed above the pedestal rotates the pedestal to continuously photograph the seedling on the pedestal. The thickness and curvature of the seedling's stem are determined from the photographed images. JP 2016-021929 A also determines the thickness of the seedling's stem from the images.

[0008] In the following Non-Patent Document 1, the internal structure of the stem of a tomato seedling is determined using CT and electron microscopes.

[0009] JP 2008-295407 A JP 2016-021929 A

[0010] American Society of Agricultural and Biological Engineers, (USA), 2016, Vol. 32(4), p. 417-427

[0011] According to Japanese Patent Application Laid-Open No. 2008-295407, Japanese Patent Application Laid-Open No. 2016-021929, and Non-Patent Document 1, the condition of seedlings can be confirmed using images, and high-quality seedlings can be used as source seedlings. However, Japanese Patent Application Laid-Open No. 2008-295407 and Japanese Patent Application Laid-Open No. 2016-021929 do not disclose specific methods for determining the thickness and curvature of seedling stems from captured images, or how numerical values ​​and thresholds for the thickness of seedling stems are determined. Furthermore, these images are limited to determining the appearance and are not capable of determining whether diseases, pests, bacteria, viruses, etc. are present inside the stems, i.e., cannot perform pathological diagnosis inside the seedlings.

[0012] Furthermore, if it takes too long to identify thin-stemmed seedlings, moisture may evaporate from the stems and leaves, causing changes to the internal structure of the stem. Therefore, it is necessary to diagnose the pathology inside the seedlings in a short time.

[0013] Non-Patent Document 1 attempts to grasp the internal structure of seedlings in 3D using CT, which is capable of tomographic imaging. However, as shown in Figures 1(a) to 1(c) of Non-Patent Document 1, the inside and surface of the seedlings are displayed as if they are crushed, and the internal structure is not clearly displayed. As such, no method has been established to distinguish the internal structure of high-quality seedlings, particularly seedlings (stems).

[0014] Furthermore, even if a method for identifying and understanding the internal structure of seedlings (stems) were established, if specialized devices and means were required for the image processing device and display means (output means) for displaying images, this could be an economic burden for seedling producers and sellers. Therefore, it is desirable that the method be executable using commonly used general-purpose devices and means (such as a personal computer) without the need for specialized devices and means.

[0015] The present invention aims to provide an image processing method, image processing device, and image processing program for seedlings that can grasp the internal structure of seedlings through images, enable the quality of seedlings to be quickly determined (diagnosed), and are versatile in their implementation.

[0016] In the present invention, multiple tomographic images of a seedling are captured using an imaging device such as a CT scanner, and the multiple two-dimensional tomographic image data are stacked to generate three-dimensional volume data including the internal structure of the subject. The data is then converted into a general-purpose format. That is, according to claim 1 of the present invention, the method includes an imaging step of capturing tomographic images of the subject, i.e., the seedling; a generation step of stacking the captured multiple two-dimensional tomographic image data in order in a direction perpendicular to the tomographic plane to generate three-dimensional volume data including the internal structure of the subject; an identification step of searching for isosurfaces within the three-dimensional volume data and identifying curved surfaces in the three-dimensional shape of the subject; and a conversion step of converting the data including the three-dimensional shape into a general-purpose format after mesh processing. According to the invention of claim 3, there is provided an image processing device for seedlings, comprising an imaging means for performing tomographic imaging of a seedling as a subject, the image processing device for seedlings performing tomographic imaging of the seedlings using the imaging means, stacking the captured multiple two-dimensional tomographic image data in a direction perpendicular to the tomographic plane to generate three-dimensional volume data including the internal structure of the subject, searching for isosurfaces within the three-dimensional volume data, identifying curved surfaces in the three-dimensional shape of the subject, and converting the data including the three-dimensional shape into a general-purpose format after mesh processing. According to the invention of claim 5, there is provided an image processing program for seedlings, comprising an imaging means for performing tomographic imaging of a seedling as a subject, the image processing program causing a computer to function as an imaging means for performing tomographic imaging of the seedling as a subject, a generating means for generating three-dimensional volume data including the internal structure of the subject by stacking the captured multiple two-dimensional tomographic image data in a direction perpendicular to the tomographic plane to generate three-dimensional volume data including the internal structure of the subject, an identifying means for searching for isosurfaces within the three-dimensional volume data and identifying curved surfaces in the three-dimensional shape of the subject, and a converting means for converting the data including the three-dimensional shape into a general-purpose format after mesh processing.

[0017] In the present invention, the internal structure of the seedlings is displayed in three dimensions, allowing the operator to quickly determine (diagnose) the quality of the seedlings using the images. In addition, the three-dimensional display can be performed using commonly used general-purpose devices and means (such as a personal computer).

[0018] 1 shows a schematic diagram of a seedling image processing device according to an embodiment of the present invention (a device that embodies a seedling image processing method and an image processing program according to an embodiment of the present invention); FIG. 2 shows a detailed schematic diagram including the seedling image processing device of FIG. 1 (a device that embodies a seedling image processing method and an image processing program); FIG. 3 shows a schematic flow diagram of a seedling image processing method according to an embodiment of the present invention; FIG. 4 shows a flow diagram of a generation process; FIG. 5 shows a flow diagram of a conversion process; FIG. 6 shows a flow diagram of a model generation process; and FIG. 7 shows an image of a three-dimensional shape including the internal structure of a displayed subject.

[0019] The method comprises an imaging step of taking tomographic images of the seedling, which is the subject; a generation step of stacking the captured two-dimensional tomographic image data in order in a direction perpendicular to the tomography to generate three-dimensional volume data including the internal structure of the subject; an identification step of searching for isosurfaces within the three-dimensional volume data and identifying curved surfaces in the three-dimensional shape of the subject; and a conversion step of converting the data including the three-dimensional shape into a general-purpose format after mesh processing.

[0020] Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. FIG. 1 shows a schematic diagram of an image processing device for seedlings (a device embodying an image processing method and an image processing program according to an embodiment of the present invention) according to an embodiment of the present invention. FIG. 2 shows a detailed schematic diagram including the image processing device (a device embodying an image processing method and an image processing program) of FIG. 1. FIG. 3 shows a schematic flow diagram of an image processing method for seedlings according to an embodiment of the present invention. As shown in FIGS. 1 and 2, an image processing device 10' for seedlings 1 (a device embodying an image processing method and an image processing program according to an embodiment of the present invention) includes imaging means 20, 10-1 for performing tomographic imaging of a seedling, which is a subject. The image processing device for seedlings performs tomographic imaging of the seedlings using the imaging means, sequentially stacks the captured multiple 2D tomographic image data 12' in a direction perpendicular to the tomographic images to generate 3D volume data including the internal structure of the subject, searches for isosurfaces within the 3D volume data, identifies curved surfaces in the 3D shape of the subject, and, after mesh processing, converts the data including the 3D shape into a general-purpose format.

[0021] Also, as shown in FIG. 3, the image processing method and image processing program by the image processing device 10' of the seedling 1 include an imaging step (S1; imaging means 20, 10-1) of taking tomographic images of the seedling 1, which is the subject; a generation step (S2; generation means 10-2) of stacking the captured two-dimensional tomographic image data 12' in a direction perpendicular to the tomography to generate three-dimensional volume data including the internal structure of the subject; an identification step (S3; identification means 10-3) of searching for an isosurface within the three-dimensional volume data and identifying a curved surface in the three-dimensional shape of the subject; and a conversion step (S4; conversion means 10-4) of converting the data including the three-dimensional shape into a general-purpose format after mesh processing.

[0022] An image processing device 10' for seedlings 1 (a device embodying an image processing method and an image processing program for seedlings according to an embodiment of the present invention) will now be described. As shown in FIG. 1 , the image processing device 10' for seedlings 1 is connected to imaging means 20, 10-1 that captures tomographic images of the subject seedling 1, and is configured to communicate with each other. As will be described in detail later, examples of the imaging means 20, 10-1 include CT and MRI. The image processing device 10' is operated, for example, by an operator at a seedling producer or distributor, but is not limited to this and may be operated by another person. Furthermore, the image processing device 10' and the imaging means 20, 10-1 may communicate directly as shown in FIG. 1 , or may communicate via a server, cloud, network, or the like (not shown). The communication is not limited to the description of FIG. 1 . Alternatively, the imaging means 20, 10-1 may include a separate image processing device (not shown), and the image processing device of the imaging means and the image processing device 10' of the present invention may be connected via a network or the like (not shown). The image processing device 10' incorporates a CPU (processor) 10'-1, which will be described later, and the image processing device is controlled by the CPU.

[0023] Referring to FIG. 2, the image processing device 10', which embodies the seedling image processing method and image processing program, includes, for example, a control unit 10, a storage unit 12, an input unit 14, and a display unit 16. Specifically, the control unit 10 is built into the image processing device 10', and the image processing device including the control unit is controlled by a CPU (processor) 10'-1 with information processing capabilities. In addition to the control unit 10 that executes various functions, the image processing device 10' also includes a storage unit 12 that stores information and is configured from a storage medium such as a flash memory, an input unit 14 that accepts input from input means 14' such as a keyboard, mouse, touch panel, or buttons, a display means (output means) 16' such as a display, and a display unit 16 that displays two-dimensional and three-dimensional images on the display 10-6. As mentioned above, the image processing device 10' may also include a communication unit (not shown) that communicates with the outside world via a network, but is not limited to this configuration.

[0024] To further describe the storage unit 12, the storage unit stores programs, data, and the like. The program is an image processing program for seedlings according to the present invention, and is read by the control unit 10 (image processing device 10′) to execute various processes, such as an imaging step S1, image processing steps S2 to S5, and a display step S6 (FIG. 3), which will be described later. The data is, for example, a two-dimensional image data set 12′, and is data that is referenced or stored as the program is executed.

[0025] The imaging means 20, 10-1 are imaging means for capturing multiple two-dimensional tomographic images, such as CT and MRI. CT uses X-rays, and MRI uses very strong magnets and electromagnetic waves to irradiate the subject from around the subject, respectively, and calculate the radiation dose transmitted through the subject to obtain a tomographic image of the subject. The imaging means 20, 10-1 are not limited to these. While the seedling 1 is positioned upright within the imaging means 20, 10-1 in FIG. 1, if the imaging means is CT or MRI, the seedling is positioned lying down within the imaging means. The position of the subject, i.e., the seedling 1, varies depending on the type of imaging means 20, 10-1. The imaging means 20, 10-1 connected to the image processing device 10' is controlled by the control unit 10 via the input unit 14', such as a keyboard or mouse, operated by an operator to control the timing of imaging. Also, as shown in FIG. 2, a two-dimensional image data set (two-dimensional tomographic image data) 12′, which is a set of multiple tomographic image data captured by the imaging means 20, 10-1, is stored in the memory unit 12 via the control unit 10.

[0026] Regarding the two-dimensional image dataset 12', when the imaging means 20, 10-1 captures tomographic images of the subject seedling 1 in the X-Y plane (a plane perpendicular to the growth direction of the seedling's stem), the images are captured at regular intervals in a direction perpendicular to the tomographic images (Z direction; mainly the growth direction of the seedling's stem). The fixed interval between images is the distance between two adjacent tomographic images (slice thickness). In the dataset 12' of multiple tomographic images captured in this manner, the individual tomographic images (two-dimensional images) are arranged in order in the direction perpendicular to the tomographic slices, that is, in the order in which they were captured. In other words, the individual tomographic images in the dataset 12' must be arranged in the order in which they were captured, and must not be arranged in any other order.

[0027] As shown in Figures 2 and 3, the control unit 10 executes an image processing program (not shown) stored in the memory unit 12, thereby functioning as (imaging means 20, 10-1), generation means 10-2, identification means 10-3, and conversion means 10-4, and sequentially performs imaging step S1, image processing steps S2 to S5, etc. That is, to perform these imaging step S1 and image processing steps S2-5, among S2-4, the control unit 10 has a computer, imaging means 20, 10-1 (S1; imaging step) that performs tomographic imaging of the seedling 1, which is the subject, generation means 10-2 (S2; generation step) that generates 3D volume data including the internal structure of the subject by sequentially stacking the captured 2D tomographic image data in a direction perpendicular to the tomographic plane, identification means 10-3 (S3; identification step) that searches for isosurfaces within the 3D volume data and identifies curved surfaces in the 3D shape of the subject, and conversion means 10-4 (S4; conversion step) that converts data including the 3D shape into a general-purpose format after mesh processing, and functions as these means. Furthermore, the control unit 10 functions as a model generation means 10-5 and display means 16', 10-6 by executing an image processing program (not shown) stored in the memory unit 12, and performs the image processing steps S2-4 of FIG. 3, followed by the image processing step S5 and the display step S6 in sequence. That is, in order to perform the image processing step S5 and the display step S6 following the imaging step S1 and the image processing steps S2 to S4, the control unit 10 further includes a model generation means 10-5 (S5; model generation step) that acquires data including the 3D shape converted by the conversion means and converts it into a 3D displayable form using vector data connecting each point cloud data included in the 3D shape data, and a display means 16', 10-6 (S6; display step) that displays the 3D shape including the internal structure of the seedling, and functions as these means. The display itself in the display step S6 is performed by the display means (output means) 16', 10-6. The model generation means 10-5 also at least sets the transparency for the 3D display.

[0028] The following mainly describes each step of the image processing method for seedlings, which is performed by the control unit 10 of the image processing device 10' for the seedlings 1. Briefly, in the imaging step S1 and image processing steps S2 to S5 in Fig. 3, a process (modeling) is performed to generate three-dimensional volume data from a data set 12' of captured two-dimensional image data.

[0029] 1. Description of Each Step 1.1 Imaging Step S1 In the imaging step S1, which is performed by the imaging means 20, 10-1, tomographic images of the seedling 1, which is the subject, are taken. The images are taken at regular intervals in a direction perpendicular to the tomographic image (Z direction; mainly the direction in which the seedling's stem grows). The images are taken, for example, by an operator using the input means 14'. The individual tomographic images (two-dimensional images) of the seedling are arranged in the order in which they are taken, and are stored in the memory unit 12 by the control unit 10 as a data set 12' of multiple tomographic images.

[0030] 1.2 Image Processing Step In the image processing step, an imaging step S1, a generation step S2, an identification step S3, and a conversion step S4 are sequentially performed on a data set 12' of multiple tomographic image data (two-dimensional image data), and general-purpose three-dimensional image data for displaying a three-dimensional image on a display means (output means) 16', 10-6 is output. Subsequently, a model generation step S5 is performed to generate a model that enables three-dimensional display.

[0031] 1.2.1 Generation Step S2 In the generation step S2, the generation means 10-2 sequentially stacks the captured 2D tomographic image data in a direction perpendicular to the tomographic planes to generate 3D volume data including the internal structure of the subject. Figure 4 shows a flow diagram of the generation step. Referring to Figure 4, the control unit 10 first clears variables (S2-1) and then acquires the full path of the directory of the dataset 12' of 2D images (tomographic images) of the subject, the seedling 1, stored in the storage unit 12 after image capture by the imaging means 20 and 10-1 (S2-2). For example, if the directory contains a mixture of 2D images that are the subject of 3D display and non-subject 2D images, the control unit 10 searches for only the target images and creates a new dataset (file list) 12' that compiles the multiple 2D images (S2-3). For example, in this embodiment, it is assumed that the two-dimensional images (tomographic images) from the imaging means 20, 10-1 are TIF (TIFF) file images, but if the directory contains images other than TIF files, only TIF files are searched for and these are used as a new data set (file list) 12'. Note that the two-dimensional images (tomographic images) from the imaging means 20, 10-1 are not limited to TIF files. If only two-dimensional images to be displayed in three dimensions are stored in the directory, step S2-3 is naturally omitted.

[0032] The control unit 10 sets an array [x, y, z, v] for the three-dimensional volume data (S2-4). x, y, and z represent the x, y, and z coordinates stored in the index of the image file of the two-dimensional image (tomographic image), and v represents, for example, the grayscale signal intensity. The image file of the two-dimensional image (tomographic image) may be either grayscale or color. In FIG. 4 , as an example, the array is designated as V0, and x, y, and z of V0 are set to 1024 ([x, y, z] = [1024, 1024, 1024]), but this is not limiting. The numerical values ​​of the array may also be set interactively. Next, the control unit 10 sets n (S2-5). Here, the control unit 10 obtains the total number n of two-dimensional images (tomographic images) included in the data set 12′ obtained in S2-3, and sets the initial value of n to 1 (S2-5).

[0033] The control unit 10 acquires the full path of the first file in the data set, which corresponds to n=1, and reads the image (image pixel data) from the image file specified by the full path (S2-6). The read image is set as I.

[0034] In steps S2-7 to S2-12, the total number n of two-dimensional image files included in the data set 12' are processed one by one. The control unit 10 sets the initial values ​​of x and y to 1 (S2-7, S2-8). The final values ​​of x and y are the values ​​set in S2-4 (in the example of FIG. 4, both x and y are 1024), and the increment values ​​of x and y are both 1.

[0035] Since the loaded image I(x,y,1) is merely image pixel data (the index value of the two-dimensional image file, i.e., coordinates), the control unit 10 stores the image I(x,y,1) in the three-dimensional volume data V0 (S2-9). As described above, the initial values ​​of x and y are each set to 1, so the initial image I is (1,1,1), and this image I is stored in V0. "n" is stored in the z field of VO because multiple two-dimensional images are arranged in the order in which they were captured in the direction perpendicular to the tomography, and the arrangement order of the two-dimensional images (tomographic images) of the arranged subject, seedling 1, corresponds to the z coordinate of the subject. In other words, the value of n (the number) in the file arrangement order corresponds to the z coordinate of the subject.

[0036] The control unit 10 repeats the storage process into the three-dimensional volume data V0 performed in S2-9 with y (increment value 1, final value 1024 (final value is an example)), x (increment value 1, final value 1024 (final value is an example)), and n (increment value 1, final value is the total number of image files (S2-5)) (S6 to S12).

[0037] By repeating this process, all of the multiple tomographic images (two-dimensional images) included in the data set 12' are layered to generate three-dimensional volume data.

[0038] 1.2.2 Identification Step S3 In the identification step S3 performed by the identification means 10-3, isosurfaces are searched for within the 3D volume data, and curved surfaces in the 3D shape of the subject are identified. That is, the control unit 10 searches for isosurfaces in the 3D volume data generated in the generation step S2 (S3), and generates and connects the isosurfaces to generate a continuous curved surface. Since isosurface processing is similar to conventional technology, only an overview will be provided here. For example, a histogram of the acquired data is created, and the data is arbitrarily binned (the histogram is divided) using the Sturgess formula or the like. If color is present, the binning results are color-coded.

[0039] Note that generating a curved surface using an isosurface is just one example, and for example, if transmittance and reflectance are guaranteed, it is possible to determine contour lines from tomographic images (two-dimensional images) and stack the images from which the contour lines are obtained to identify a three-dimensional shape. In other words, other methods may be used to generate a curved surface and identify a three-dimensional shape.

[0040] 1.2.3 Conversion Step S4 In the conversion step S4 performed by the conversion means 10-4, after mesh processing, the data containing the three-dimensional shape is converted into a general-purpose format. FIG. 5 shows a flow diagram of the conversion step. Referring to FIG. 5, the control unit 10 first approximates and interpolates the curved surface using mesh processing (S4-1). Then, the data containing the three-dimensional shape is converted into a general-purpose format (S4-2). The control unit 10 may temporarily store the converted data in the storage unit 12. A general-purpose format (file format) may be, for example, the Standard Triangulated Language (STL) format, but the file format is not limited to this. The STL format is one file format for saving data representing three-dimensional shapes. Another example of a data format is a binary format, but this is not limited thereto. The mesh processing is similar to that of conventional techniques, so details will not be described here. The mesh shape may be any shape, such as a triangle or a rectangle.

[0041] 1.2.4 Model Generation Step S5 The data output in the conversion step S4 merely contains coordinates representing the three-dimensional curved surface (three-dimensional shape) of the seedling 1. In the model generation step S5, performed by the model generation means 10-5, data containing the three-dimensional shape converted in the conversion step S4 is acquired and converted into a three-dimensional displayable form using vector data connecting the point cloud data contained in the three-dimensional shape data. That is, in the model generation step S5, the model generation means 10-5 generates a three-dimensional rendering image. Figure 6 shows a flow diagram of the model generation step. Referring to Figure 6, the control unit 10 first clears variables (S5-1) and sets up, for example, a parallel engine for high-speed processing (S5-2). Note that the parallel engine is technically configured for parallel processing to improve processing speed and is not required. It goes without saying that processing methods other than parallelization may be used for high-speed processing. However, the following description assumes that high-speed processing is performed using parallel processing.

[0042] The control unit 10 acquires the data in a general-purpose format converted and output in the conversion step S4 through parallel processing (S5-3), and reads the point cloud data (specifically, the vertices and faces of the three-dimensional shape) contained in the data (S5-4).

[0043] The control unit 10 then constructs a curved surface by connecting points and vectors based on the loaded point cloud data and the vector data linking the point cloud data to connect the vertices (S5-5). The control unit 10 sets at least the transparency (parameter, also called opacity) of the curved surface (S5-6). While setting the transparency is essential, other settings necessary for displaying the three-dimensional shape, such as the edge color, the direction of light on the surface, the intensity of ambient light, color, aspect ratio, etc., are also appropriately performed. These settings (including changes) may also be performed interactively.

[0044] 1.3. Display Step S6 In the display step S6, which is performed by the display means (output means) 16, a three-dimensional shape including the internal structure of the seedling is displayed. The control unit 10 displays the three-dimensional shape of the seedling 1 on the display means (output means) 16', 10-6 based on the data rendered in the model generation step. Figure 7 shows an image of the three-dimensional shape including the internal structure of the displayed subject. Note that Figure 7 is merely a portion (screenshot) of a rotatable image of the three-dimensional shape. While Figure 7 would normally be a grayscale image (or an image with transparency), please note that it is not an accurate grayscale image due to the application regulations. More specifically, Figure 7 shows the three-dimensional display obtained by performing image processing steps S2 to S5 and display step S6 on a dataset 12' of cross-sectional images of the stem of the seedling 1. The three-dimensional image displayed in the display step S6 can be rotated in the x, y, and z directions on the display means 16', 10-6 by dragging with the input means 14' (mouse), and it is also possible to enter the internal structure and display and operate it as if looking at the internal structure from the inside. In the example of Figure 7, the multiple thin tubes displayed inside are vascular bundles. As shown in Figure 7, the vascular bundles are not interrupted in the middle or displayed as if they are integrated into the surrounding tissue and crushed, but are displayed clearly and semi-transparently.

[0045] 2. Effects In the image processing method and program for seedling 1, and the image processing device 10' that embodies them, in the generation step S2, two-dimensional image data (files) are read one by one to generate three-dimensional volume data, and isosurface processing (identification step S3) and mesh processing (conversion step S4) are performed successively in this order, thereby retaining the three-dimensional shape of the seedling's internal structure contained in the three-dimensional volume data as data (coordinates). In the subsequent model generation step S5, transparency processing is performed while retaining the data (coordinates) related to the internal structure of the three-dimensional shape of the seedling 1. Therefore, in the display step S6, the internal structure of the seedling 1, i.e., the three-dimensional shape of the vascular bundles and other components, is displayed semi-transparently without being distorted, as shown in FIG. 7.

[0046] In the conversion step S4, the three-dimensional volume data is converted into a general-purpose format (e.g., an STL file), thereby achieving data compatibility. That is, by converting the three-dimensional volume data into a general-purpose format (e.g., an STL file) in the conversion step S4 and outputting it once, the data can be transferred to other image processing devices 10', display means 16', 10-6, etc. Furthermore, a binary format also reduces the data size. In particular, in the generation step S2, only the array [x, y, z, v] for the three-dimensional volume data (x, y, z are the x, y, z coordinates stored in the index of the image file of the two-dimensional image (tomographic image), and v is the grayscale signal intensity) is required, and no other parameters are required (S2-4). This also makes it possible to achieve data compatibility.

[0047] The more versatile the image file format, the more inconsistent the display will be on the image processing device 10' (particularly the control unit 10) and the display means 16', 10-6 when displayed in 3D. Therefore, by performing the model generation step S5, a unified 3D display becomes possible regardless of the image processing device 10' (particularly the control unit 10).

[0048] The three-dimensional image displayed in the display step S6 can be rotated in the x, y, and z directions, and the subject can be displayed from any angle, for example, by inputting instructions using a mouse, which is one of the input means 14'. Furthermore, the display means 16', 10-6, can also be used to enter the internal structure of the seedling 1, allowing for display and operation as if peering into the internal structure from the inside. This can assist in the diagnosis of workers who are unfamiliar with diagnosis (workers at seedling producers and distributors).

[0049] The internal structure of the seedling 1 can be displayed in three dimensions using a commonly used device (image processing device 10') and a display (display means 16', 10-6). The isosurface processing (identification step S3) and mesh processing (conversion step S4) are performed automatically without requiring any particular time. The model generation step S5 is also performed automatically, and the image is automatically displayed after the display step S6. Therefore, even an operator unfamiliar with image processing and other operations can easily display the image in three dimensions, and can quickly determine (diagnose) the quality of the seedling, i.e., whether or not there is any disease, pests, or the like in the vascular bundles, by referring to the displayed image.

[0050] By using an imaging means 20 such as a CT scanner that can capture images of the internal structure, it is possible to display and observe the detailed internal structure of the seedling 1, and to measure how the fusion (taking root) of the scion and rootstock progresses. Understanding the state of fusion (taking root) can shorten the number of days for curing and improve yields. It is also possible to observe how bacterial and viral diseases, pests, and other factors affect the internal structure of the seedling, including the vascular bundles. This can stabilize the quality of grafted seedlings and contribute to stabilizing Japan's food situation.

[0051] 3. Other The above-described examples are for the purpose of explaining the present invention and are not intended to limit the present invention in any way. It goes without saying that all modifications and alterations within the technical scope of the present invention are encompassed by the present invention. For example, the imaging means may be X-ray, MRI, echo, light, or the like. Furthermore, data may be output from a combination of different imaging means, such as a combination of CT and MRI. In other words, regardless of the type of imaging means, as long as the output (image data, etc.) from the imaging means can be electrically converted, the seedling image processing method of the present invention can be applied.

[0052] 4. Supplementary Note 4.1 Image Processing Method for Seedlings This method includes an imaging step of taking tomographic images of a seedling as a subject, a generation step of generating three-dimensional volume data including the internal structure of the subject by sequentially stacking the captured two-dimensional tomographic image data in a direction perpendicular to the tomographic plane, an identification step of searching for an isosurface within the three-dimensional volume data and identifying curved surfaces in the three-dimensional shape of the subject, and a conversion step of converting data including the three-dimensional shape into a general-purpose format after mesh processing. The method further includes a model generation step of acquiring the data including the three-dimensional shape converted in the conversion step and converting it into a form that can be displayed in three dimensions using vector data connecting each point cloud data included in the three-dimensional shape data, and a display step of displaying the three-dimensional shape including the internal structure of the seedling, where at least transparency for the three-dimensional display is set in the model generation step. 4.2 Seedling Image Processing Device This seedling image processing device includes an imaging means for capturing tomographic images of a seedling as a subject, and is characterized in that it captures tomographic images of the seedling using the imaging means, sequentially stacks the captured multiple 2D tomographic image data in a direction perpendicular to the tomographic plane to generate 3D volume data including the internal structure of the subject, searches for isosurfaces within the 3D volume data, identifies curved surfaces in the 3D shape of the subject, and, after mesh processing, converts the data including the 3D shape into a general-purpose format. The seedling image processing device further includes a display means for displaying the 3D shape, and is characterized in that it acquires data including the converted 3D shape, converts it into a 3D displayable form using vector data connecting each point cloud data included in the 3D shape data, sets at least the transparency for the 3D display, and displays the 3D shape including the internal structure of the seedling on the display means. 4.3 Seedling Image Processing Program In an image processing program for seedlings that includes an imaging means for taking tomographic images of the seedlings that are the subject, a computer functions as an imaging means for taking tomographic images of the seedlings that are the subject, a generation means for sequentially stacking the captured two-dimensional tomographic image data in a direction perpendicular to the tomography to generate three-dimensional volume data including the internal structure of the subject, an identification means for searching for isosurfaces within the three-dimensional volume data and identifying curved surfaces in the three-dimensional shape of the subject, and a conversion means for converting data including the three-dimensional shape into a general-purpose format after mesh processing.The system further includes a model generation means for acquiring data including the three-dimensional shape converted by the conversion means and converting it into a three-dimensional displayable form using vector data connecting each point cloud data included in the three-dimensional shape data, and a display means for displaying the three-dimensional shape including the internal structure of the seedling, and the model generation means at least sets the transparency for the three-dimensional display.

[0053] Although the invention is typically applied to agricultural imaging, it can also be applied to medical imaging to capture fine internal structures.

[0054] 1 Seedling 10 Control unit 10-1, 20 Imaging means 10-2 Generation means 10-3 Identification means 10-4 Conversion means 10-5 Model generation means 10-6, 16' Display means (output means) 10' Seedling image processing device 12' Two-dimensional image data set (two-dimensional tomographic image data) S1 Imaging process S2 Generation process S3 Identification process S4 Conversion process S5 Model generation process S6 Display process

Claims

1. An image processing method for seedlings, comprising: an imaging step of capturing tomographic images of the seedling as a subject; a generation step of generating three-dimensional volume data including the internal structure of the subject by stacking the captured two-dimensional tomographic image data in a direction perpendicular to the tomographic images; an identification step of searching for isosurfaces within the three-dimensional volume data and identifying curved surfaces in the three-dimensional shape of the subject; and a conversion step of converting the data including the three-dimensional shape into a general-purpose format after mesh processing.

2. The seedling image processing method according to claim 1, further comprising: a model generation step of acquiring data including the three-dimensional shape converted in the conversion step, and converting the data into a three-dimensional displayable form using vector data connecting each point cloud data included in the three-dimensional shape data; and a display step of displaying the three-dimensional shape including the internal structure of the seedling, wherein the model generation step at least sets the transparency for three-dimensional display.

3. A seedling image processing device equipped with an imaging means for capturing tomographic images of a seedling as a subject, the seedling image processing device: captures tomographic images of the seedling using the imaging means; generates three-dimensional volume data including the internal structure of the subject by stacking multiple captured two-dimensional tomographic image data in a direction perpendicular to the tomographic plane; searches for an isosurface within the three-dimensional volume data, identifies curved surfaces in the three-dimensional shape of the subject; and converts the data including the three-dimensional shape into a general-purpose format after mesh processing.

4. The seedling image processing device according to claim 3, further comprising a display means for displaying the three-dimensional shape, acquiring data including the converted three-dimensional shape, converting the data into a three-dimensional displayable form using vector data connecting each point cloud data included in the three-dimensional shape data, setting at least the transparency for the three-dimensional display, and displaying the three-dimensional shape including the internal structure of the seedling on the display means.

5. An image processing program for seedlings having an imaging means for taking tomographic images of the subject, the program causing a computer to function as: an imaging means for taking tomographic images of the subject, the seedling; a generating means for generating three-dimensional volume data including the internal structure of the subject by stacking multiple captured two-dimensional tomographic image data in a direction perpendicular to the tomographic plane; an identifying means for searching for an isosurface within the three-dimensional volume data and identifying curved surfaces in the three-dimensional shape of the subject; and a converting means for converting data including the three-dimensional shape into a general-purpose format after mesh processing.

6. A seedling image processing program as claimed in claim 5, further comprising: a model generation means for acquiring data including the three-dimensional shape converted by the conversion means, and converting the data into a three-dimensional displayable form using vector data connecting each point cloud data included in the three-dimensional shape data; and a display means for displaying the three-dimensional shape including the internal structure of the seedling, wherein the model generation means at least sets the transparency for the three-dimensional display.

Images