Grain thinning support device, grain thinning support system, grain thinning support method and program

Abstract

To provide a grapes picking support device capable of counting the number of grapes attached to a grape bunch, a grapes picking support system, a grapes picking support method and a program.SOLUTION: A grapes picking support device includes: a moving image acquisition unit 211 for acquiring moving image data obtained by photographing while moving an imaging unit 22 around a grape bunch; an image selection unit 212 for selecting a pair of temporally adjacent still images data; a bunch image segmentation unit 213 for segmenting a bunch image about each of the pair of still images data; a same bunch image specification unit 217 for specifying bunch images corresponding to the same grape bunch on the basis of the coincidence of at least one bunch image included in each of the segmented pair of still images data; a number of grapes calculation unit 218 for calculating the number of grapes of grape included in each bunch image corresponding to the same grape bunch; and a number of grapes determination unit 219 for determining the average number of grapes calculated about all of the selected pair of still images data as the number of grapes attached to a grape bunch.SELECTED DRAWING: Figure 3

Description

Technical Field

[0001] The present invention relates to a berry thinning support device, a berry thinning support system, a berry thinning support method, and a program. Background Art

[0002] A measuring device has been proposed that counts the number of grape berries included in a grape bunch by using captured images of the bunch from multiple directions obtained by imaging the grape bunch multiple times while rotating the grape bunch by about 120 to 150 degrees, and guides berries to be thinned out from the bunch through image analysis so that the density of grape berries within the bunch is uniform (see, for example, Patent Document 1). This measuring device sequentially performs, on a plurality of captured images, specification of a pixel range corresponding to a grape bunch that is a measurement object in the captured image, and counting of the number of berries within the pixel range. Then, the measuring device grasps the rotation angle of the bunch from the movement amount of berries in the images, and prevents the initially counted berry from being counted again even if the berry moves to another position due to rotation. Prior Art Literature Patent Literature

[0003] Patent Document 1 Japanese Unexamined Patent Application Publication No. 2020-60505 Summary of the Invention Problem to be Solved by the Invention

[0004] However, in the case of the measuring device described in Patent Document 1, when one captured image includes images of a plurality of grape bunches, it is not possible to select an image of a grape bunch to be counted from among the images of the plurality of grape bunches. For this reason, for example, there is a risk that images of a plurality of grape bunches may be recognized as an image of the same grape bunch, resulting in output of an incorrect count value.

[0005] This invention has been made in view of the above-mentioned reasons, and aims to provide a grape thinning support device, grape thinning support system, grape thinning support method, and program that can accurately count the number of grapes on a bunch. [Means for solving the problem]

[0006] To achieve the above objective, the grain thinning support device according to the present invention is Imaging unit, A video acquisition unit that acquires moving image data obtained by moving the aforementioned imaging unit around a bunch of grapes, An image selection unit that sequentially selects a pair of still image data that are adjacent in time from the still image data of multiple frames that constitute the aforementioned moving image data, starting from the first still image data in time. Each time a pair of temporally adjacent still image data is selected, a cluster image extraction unit extracts a cluster image from each of the pair of still image data that shows at least one cluster of grapes, A unit that identifies a cluster image corresponding to the same grape cluster based on the degree of similarity of at least one cluster image contained in each of the extracted pair of still image data, A grape count calculation unit calculates the number of grapes contained in each bunch image corresponding to the same bunch of grapes contained in each of the extracted still image data, The system includes a grape count determination unit that determines the number of grapes on a bunch of grapes by calculating the average value of the number of grapes calculated for all pairs of still image data sequentially selected from multiple still image data frames. [Effects of the Invention]

[0007] According to the present invention, the identical bunch image identification unit identifies bunch images corresponding to the same bunch of grapes based on the degree of agreement of at least one bunch image contained in each of the extracted pair of still image data, and the grape count calculation unit calculates the number of grapes contained in each bunch image corresponding to the same bunch of grapes contained in each of the extracted pair of still image data. Then, the grape count determination unit determines the number of grapes on the bunch as the average value of the grape counts calculated for all pairs of still image data sequentially selected from multiple frames of still image data. As a result, even if the still image data constituting the moving image data contains multiple bunch images, the bunch image corresponding to the same bunch of grapes to be used for grape count calculation can be identified from among them, and the number of grapes contained in that bunch can be determined, so the number of grapes on the bunch can be counted with high accuracy. [Brief explanation of the drawing]

[0008] [Figure 1] This is a schematic diagram of the grain thinning support device according to Embodiment 1 of the present invention. [Figure 2] This is a block diagram showing the hardware configuration of the grain thinning support device according to Embodiment 1. [Figure 3] This is a block diagram showing the functional configuration of the grain thinning support device according to Embodiment 1. [Figure 4] This is an explanatory diagram of the operation of the grain thinning support device according to Embodiment 1. [Figure 5] This flowchart shows an example of the flow of the thinning support process performed by the thinning support device according to Embodiment 1. [Figure 6] This is a schematic diagram of the grain thinning support system according to Embodiment 2 of the present invention. [Figure 7] This is a block diagram showing the hardware configuration of the grain thinning support system according to Embodiment 2. [Figure 8] This is a block diagram showing the functional configuration of the grain thinning support device according to Embodiment 2. [Figure 9] This is a block diagram showing the functional configuration of the cloud server according to Embodiment 2. [Figure 10]This figure shows an example of the information stored by the particle position history storage unit according to Embodiment 2. [Figure 11] This figure shows an example of a grain thinning position estimation model used by the grain thinning position estimation unit according to Embodiment 2. [Figure 12] This is a sequence diagram illustrating the operation of the grain thinning support system according to Embodiment 2. [Figure 13] This flowchart shows an example of the flow of the thinning support process performed by the thinning support device according to Embodiment 2. [Figure 14] This flowchart shows an example of the flow of the grain thinning position notification process executed by the cloud server according to Embodiment 2. [Modes for carrying out the invention]

[0009] (Embodiment 1) Hereinafter, a grape thinning support device according to an embodiment of the present invention will be described with reference to the drawings. The grape thinning support device according to this embodiment comprises an imaging unit, a video acquisition unit that acquires video data obtained by moving the imaging unit around a bunch of grapes, an image selection unit, a bunch image extraction unit, a same bunch image identification unit, a grape number calculation unit, and a grape number determination unit. Here, the image selection unit sequentially selects a pair of still image data that are adjacent in time from a plurality of still image data frames that constitute the video data, starting from the first still image data in time. Each time a pair of still image data that are adjacent in time is selected, the bunch image extraction unit extracts a bunch image of the portion showing at least one bunch of grapes from each of the pair of still image data. The same bunch image identification unit identifies a bunch image corresponding to the same bunch of grapes based on the degree of agreement of at least one bunch image contained in each of the extracted pair of still image data. The grape number calculation unit calculates the number of grapes contained in each bunch image corresponding to the same bunch of grapes contained in each of the extracted pair of still image data. The grape count determination unit determines the number of grapes on a bunch of grapes by taking the average value calculated for all pairs of still image data sequentially selected from multiple still image data frames.

[0010] The berry thinning support apparatus 2 according to the present embodiment is used by grape farmers and the like, and as shown in Fig. 1, comprises a glass unit 20, and a control unit 25 connected to the glass unit 20 via a cord 24. The glass unit 20 has a so-called glasses-shaped configuration, and can be worn by a grape farmer during berry thinning work. The control unit 25 has a small housing that can be stored, for example, in a pocket of a worker's clothes or the like. The glass unit 20 includes an imaging unit 22 configured to capture an image of the area in front of the worker, an on-glass display unit 23, and a frame 21 that supports the imaging unit 22 and the on-glass display unit 23 and is worn on the worker's head. The on-glass display unit 23 includes, for example, a transparent or semi-transparent liquid crystal panel, an organic EL (Electro-Luminescence) panel, or the like. A worker can check information displayed on the on-glass display unit 23 while checking the front through the on-glass display unit 23. Specifically, the worker can check the number of berry grains to be thinned displayed on the on-glass display unit 23 while visually recognizing the grapes to be thinned through the on-glass display unit 23.

[0011] The imaging unit 22 is configured to capture an image of grapes that are targets of berry thinning. The imaging unit 22 includes, for example, an ordinary visible light camera. Note that the imaging unit 22 may be a near-infrared camera, a multispectral camera, a specific wavelength imaging camera, a thermal imaging camera, or the like. The imaging unit 22 also includes an imaging element such as a CCD (Charge Coupled Device) sensor or a CMOS (Complementary Metal Oxide Semiconductor) sensor. The imaging unit 22 generates image information obtained by converting electrical signals obtained by photoelectrically converting light received by the imaging element into digital data, and transfers the image information to the control unit 25.

[0012] As shown in Figure 2, the control unit 25 includes a CPU (Central Processing Unit) 201, a main storage unit 202, an auxiliary storage unit 203, a display interface (hereinafter referred to as "I / F") 204, an input unit 205, an imaging I / F 207, an audio I / F 208, and a bus 209 that connects these components to each other. The main storage unit 202 includes a volatile memory such as RAM (Random Access Memory), and is used as a work area for the CPU 201. The auxiliary storage unit 203 is a non-volatile memory such as a semiconductor flash memory, and stores programs for the CPU 201 to execute various processes. The display I / F 204 is a display device such as a display that converts image information transferred via the bus 209 into a signal for causing an on-glass display unit 23 to display an image, and outputs the signal to the on-glass display unit 23. The input unit 205 is a switch such as a key, a switch, a dial, or a touch pad disposed on the housing of the control unit 25. An operator can turn power on or off for the berry harvesting support device 2 by operating the input unit 205. The input unit 205 transfers operation information corresponding to the content of an operation performed by the operator on the input unit 205 to the CPU 201.

[0013] The CPU 201 reads the program stored in the auxiliary storage unit 203 into the main storage unit 202 and executes it, so as shown in Figure 3, it functions as a video acquisition unit 211, an image selection unit 212, a bunch image extraction unit 213, a grape position estimation unit 214, a distance calculation unit 215, a counting target identification unit 216, a same bunch image identification unit 217, a grape number calculation unit 218, a grape number determination unit 219, a display control unit 220, a grape number determination unit 223, and a notification unit 224. Furthermore, the auxiliary storage unit 203 shown in Figure 2 has a video image storage unit 231 that stores video image data obtained by imaging unit 22, a model storage unit 232, and a grape position storage unit 233 that stores grape position information indicating the estimated positions of grapes, as shown in Figure 3. In addition, the main storage unit 202 shown in Figure 2 has a selected image storage unit 229 that temporarily stores a pair of still image data selected from a plurality of still image data that constitute the video image data, as shown in Figure 3. When a new pair of still image data is stored in the selected image storage unit 229, it erases the pair of still image data that was stored immediately before.

[0014] The model memory unit 232 stores grape position estimation model information, which indicates a grape position estimation model used to estimate the positions of multiple grapes that make up a bunch of grapes. Here, the grape position estimation model is, for example, a CNN (Convolutional Neural Network), and when bunch image information showing a grape bunch image is input, it outputs the relative position coordinates of the center positions of each of the multiple grapes in the bunch image from the bunch image information. Here, the bunch image information includes, for example, the relative position coordinates of each of the multiple pixels that make up the bunch image, and the pixel value at each position coordinate. The grape position estimation model information indicates a trained grape position estimation model.

[0015] The video acquisition unit 211 acquires video data from the imaging unit 22, which is obtained when an operator moves the imaging unit 22 around the bunch of grapes to be thinned. The video acquisition unit 211 stores the acquired video data in the video data storage unit 231. The image selection unit 212 sequentially selects pairs of still image data that are adjacent in time from the multiple still image data frames that make up the video data stored in the video data storage unit 231, starting from the first still image data in time. The image selection unit 212 sequentially stores the selected pairs of still image data in the selected image storage unit 229.

[0016] Each time a pair of temporally adjacent still image data is selected and stored in the selected image storage unit 229, the cluster image extraction unit 213 extracts a cluster image from each of the pair of still image data that shows at least one cluster of grapes. Here, the cluster image extraction unit 213 identifies cluster images RA[k] and RA[k+1] corresponding to the grape cluster from the still image data corresponding to a pair of temporally adjacent frames (frame k and frame k+1) included in the moving image data, for example, as shown in Figure 4. The cluster image extraction unit 213 notifies the grape position estimation unit 214 of the cluster image information indicating the identified cluster images RA[k] and RA[k+1] corresponding to the grape cluster.

[0017] The grape position estimation unit 214 estimates the position of each grape in the still image data for the bunch image extracted from the selected pair of still image data. Specifically, the grape position estimation unit 214 uses the grape position estimation model indicated by the grape position estimation model information stored in the model storage unit 232 to estimate the relative position coordinates of each of the multiple grapes in the bunch image data and the center position coordinates of the grape bunch from the bunch image information notified by the bunch image extraction unit 213, and generates position coordinate information indicating the estimated relative position coordinates and the center position coordinates of the grape bunch. The grape position estimation unit 214 notifies the distance calculation unit 215 and the same bunch image identification unit 217 of the generated position coordinate information.

[0018] The distance calculation unit 215 estimates the distance between the center position of the grape cluster and the center position of the image, and calculates the Euclidean distance between each and the center position of the still image containing the corresponding cluster image. The distance calculation unit 215 calculates the Euclidean distance between the relative position of the center position of the grape cluster indicated by the position coordinate information notified by the berry position estimation unit 214 and the center position of the still image. Here, if the still image contains multiple cluster images, the distance calculation unit 215 calculates the Euclidean distance between the relative position of the center position of the grape cluster corresponding to each of the multiple cluster images and the center position of the still image. Then, the distance calculation unit 215 calculates the average value of the calculated Euclidean distances and notifies the counting target identification unit 216 of the distance information indicating the calculated average value.

[0019] The counting target identification unit 216 identifies the bunch image with the smallest average value of the calculated Euclidean distance as the bunch image corresponding to the bunch of grapes to be counted. When the distance calculation unit 215 notifies the counting target identification unit 216 of distance information corresponding to each of the multiple bunch images, it compares the average value of the Euclidean distance indicated by each of the notified distance information and identifies the bunch image with the smallest average value as the bunch image corresponding to the bunch of grapes to be counted. The counting target identification unit 216 notifies the same bunch image identification unit 217 of the bunch image information indicating the identified bunch image.

[0020] The identical bunch image identification unit 217 identifies bunch images corresponding to the same bunch of grapes based on the degree of agreement of at least one bunch image contained in each pair of still image data extracted by the bunch image extraction unit 213. Specifically, the identical bunch image identification unit 217 identifies a bunch image as corresponding to the same bunch of grapes if the degree of agreement of the relative arrangement of grapes indicated by the position coordinate information corresponding to the bunch image contained in each pair of still image data notified by the grape position estimation unit 214 is equal to or greater than a preset standard agreement degree. For example, as shown in Figure 4, the identical bunch image identification unit 217 identifies a bunch image as corresponding to the same bunch of grapes if the degree of agreement of the relative arrangement of grapes in the matching portion Ad of bunch images RA[k] and RA[k+1] contained in still image data corresponding to a pair of temporally adjacent frames (frame k and frame k+1) is equal to or greater than the standard agreement degree. Returning to Figure 3, the identical bunch image identification unit 217 then stores the position coordinate information corresponding to the identified bunch image in the grape position storage unit 233.

[0021] The grape count calculation unit 218 calculates the number of grapes in a bunch based on the bunch images corresponding to the same bunch of grapes contained in each of the extracted still image data. Specifically, the grape count calculation unit 218 counts the number of relative position coordinates indicated by the position coordinate information corresponding to the bunch image stored in the grape position storage unit 233, and calculates the number of grapes based on the counted number of relative position coordinates. The grape count calculation unit 218 notifies the grape count determination unit 219 of the calculated grape count information.

[0022] The grape count determination unit 219 determines the number of grapes in a bunch of grapes by calculating the average value of the number of grapes calculated for all pairs of still image data sequentially selected from multiple frames of still image data. The grape count determination unit 219 calculates the average value of the number of grapes indicated by the grape count information corresponding to the same bunch of grapes in multiple pairs of still image data notified by the grape count calculation unit 218, and determines the calculated average value as the number of grapes in the bunch. The grape count determination unit 219 notifies the display control unit 220 and the grape count determination unit 223 of the grape count information indicating the determined number of grapes.

[0023] The display control unit 220 forms a grape count notification image information, which represents the number of grapes to be counted, based on the grape count information notified from the grape count determination unit 219, and transmits it to the display I / F 204. As a result, the grape count notification image is displayed on the on-glass display unit 23.

[0024] The grape count determination unit 223 determines whether the number of grapes determined by the grape count determination unit 219 is less than or equal to a preset standard number of grapes. The grape count determination unit 223 determines whether the number of grapes indicated by the grape count information notified by the grape count determination unit 219 is less than or equal to the aforementioned standard number of grapes. If the grape count determination unit 223 determines that the number of grapes indicated by the grape count information is less than or equal to the aforementioned standard number of grapes, it notifies the notification unit 224 of a notification command information instructing it to notify the worker of this fact by voice. When the notification unit 224 receives the notification command information, it generates preset voice information and transfers it to the voice interface 208. As a result, the speaker 26 outputs a notification sound corresponding to the voice information.

[0025] Next, the grain thinning support process performed by the grain thinning support device 2 according to this embodiment will be described with reference to Figure 5. This grain thinning support process is started, for example, when power is turned on to the grain thinning support device 2. In parallel with this grain thinning support process, the video acquisition unit 211 is assumed to be performing the operation of acquiring video image data obtained by the imaging unit 22 and sequentially storing it in the video image storage unit 231. First, the image selection unit 212 determines whether or not a preset grain count display time has arrived (step S101). This grain count display time is set to arrive, for example, at intervals of several milliseconds. In this case, the grain count notification image will be updated sequentially at intervals of several tens of milliseconds. Here, as long as the image selection unit 212 determines that the aforementioned grain count display time has not yet arrived (step S101: No), the process in step S101 is repeatedly executed. On the other hand, when the image selection unit 212 determines that the aforementioned grain count display time has arrived (step S101: Yes), it selects a pair of still image data that are adjacent in time from the still image data of multiple frames that constitute the moving image data stored in the moving image storage unit 231, and sequentially stores the selected pair of still image data in the selected image storage unit 229 (step S102).

[0026] Next, the cluster image extraction unit 213, once a pair of temporally adjacent still image data has been selected and stored in the selected image storage unit 229, extracts a cluster image from each of the pair of still image data that shows at least one cluster of grapes (step S103). Subsequently, the grain position estimation unit 214 estimates the position of the individual grapes and the center position of the grape cluster in the still image data, based on the cluster image extracted from the selected pair of still image data (step S104).

[0027] Subsequently, the distance calculation unit 215 calculates the Euclidean distance between the estimated center position of the grape cluster and the center position of the still image containing the corresponding cluster image, and calculates the average value of the calculated Euclidean distances (step S105). Next, the counting target identification unit 216 identifies the cluster image with the smallest average value of the calculated Euclidean distances as the cluster image corresponding to the grape cluster for which the number of grapes is to be counted (step S106).

[0028] Next, the identical bunch image identification unit 217 identifies bunch images corresponding to the same bunch of grapes based on the degree of agreement of at least one bunch image contained in each of the pair of still image data extracted by the bunch image extraction unit 213 (step S107). Then, the grape count calculation unit 218 calculates the number of grapes contained in the bunch of grapes based on the bunch images corresponding to the same bunch of grapes contained in each of the extracted pair of still image data (step S108). Next, the grape count determination unit 219 determines the average value of the grape counts calculated for all pairs of still image data sequentially selected from multiple frames of still image data as the number of grapes contained in the bunch of grapes (step S109). Subsequently, the display control unit 220 forms grape count notification image information showing a grape count notification image representing the number of grapes to be counted based on the grape count information notified by the grape count determination unit 219 and displays it on the on-glass display unit 23 (step S110).

[0029] Subsequently, the grape count determination unit 223 determines whether the number of grapes determined by the grape count determination unit 219 is less than or equal to a preset standard number of grapes (step S111). If the grape count determination unit 223 determines that the number of grapes determined by the grape count determination unit 219 is greater than the aforementioned standard number of grapes (step S111: No), the process in step S101 is executed again. On the other hand, if the grape count determination unit 223 determines that the number of grapes determined by the grape count determination unit 219 is less than or equal to the aforementioned standard number of grapes (step S111: Yes), it notifies the notification unit 224 of a notification command information instructing it to notify the worker of this fact by voice. When the notification unit 224 receives the notification command information, it generates preset voice information and outputs a notification sound corresponding to the generated voice information from the speaker 26 (step S112). Next, the process in step S101 is executed again.

[0030] As described above, according to the grape thinning support device 2 of this embodiment, the same bunch image identification unit 217 identifies bunch images corresponding to the same grape bunch based on the degree of agreement of at least one bunch image contained in each of the extracted pair of still image data. The grape count calculation unit 218 calculates the number of grapes contained in the bunch image based on the bunch image corresponding to the same grape bunch contained in each of the extracted pair of still image data. The grape count determination unit 219 then determines the number of grapes on the grape bunch as the average value of the grape counts calculated for all pairs of still image data sequentially selected from multiple frames of still image data. As a result, even if the still image data constituting the moving image data contains multiple bunch images, the bunch image corresponding to the same grape bunch to be used for grape count calculation can be identified from among them, and the number of grapes on that bunch can be determined, thus enabling accurate counting of the number of grapes on the bunch.

[0031] (Embodiment 2) The grape thinning support system according to this embodiment generates a grape thinning position estimation model using the history of relative position coordinates of grapes from the start of thinning to the most recent thinning, and has the function of estimating the position of grapes to be thinned from the position of grapes identified in the bunch image of the grape bunch before thinning using the generated grape thinning position estimation model. The grape thinning support system according to this embodiment comprises a grape thinning support device 2002 and a cloud server 1, as shown in Figure 6. In Figure 6, components similar to those in Embodiment 1 are denoted by the same reference numerals as in Figure 1. Furthermore, a local network NW2 is constructed at the grape farm, and an access point 82 connected to the local network NW2 and capable of wireless communication with the grape thinning support device 2002, and a broadband router (hereinafter referred to as "BBR") 81 connected to the local network NW2 and capable of communicating with the cloud server 1 via the wide area network NW1 are installed. The wide area network NW1 is, for example, the internet. Furthermore, the local network NW2 is, for example, a wireless LAN (Local Area Network).

[0032] The grape thinning support device 2002 comprises a glass unit 20 as described in Embodiment 1, and a control unit 2025 connected to the glass unit 20 via a cord 24. The control unit 2025 has a small housing that can be stored, for example, in a pocket of clothing by an operator, similar to the control unit 25 described in Embodiment 1. The operator can then visually inspect the grapes to be thinned through the on-glass display unit 23 and confirm not only the number of grapes to be thinned, but also information on the position of the next candidate grape to be thinned, as displayed on the on-glass display unit 23.

[0033] As shown in Figure 7, the control unit 2025 includes a CPU 201, a main memory unit 202, an auxiliary memory unit 203, a display interface 204, an input unit 205, a wireless module 206, an imaging interface 207, an audio interface 208, and a bus 209 connecting these to each other. In Figure 7, components similar to those in Embodiment 1 are denoted by the same reference numerals as in Figure 2. The wireless module 206 communicates with the access point 82 using a communication method compliant with, for example, the IEEE 802.11 wireless communication standard, and transmits information transferred from the bus 209 to the access point 82, and sends information received from the access point 82 to the bus 209.

[0034] The CPU 201 reads the program stored in the auxiliary storage unit 203 into the main storage unit 202 and executes it, so that, as shown in Figure 8, it functions as a video acquisition unit 211, an image selection unit 212, a bunch image extraction unit 213, a grain position estimation unit 214, a distance calculation unit 215, a counting target identification unit 216, a same bunch image identification unit 2217, a grain number calculation unit 218, a grain number determination unit 219, a display control unit 220, a position coordinate notification unit 2221, a grain thinning position acquisition unit 2222, a grain number determination unit 223, and a notification unit 224. Note that in Figure 8, components similar to those in Embodiment 1 are denoted by the same reference numerals as in Figure 3.

[0035] The cluster image identification unit 2217 stores the position coordinate information corresponding to the identified cluster image in the grape position storage unit 233 and notifies the position coordinate information corresponding to the identified cluster image to the position coordinate notification unit 2221. The position coordinate notification unit 2221 transmits the position coordinate information notified by the cluster image identification unit 2217 to the cloud server 1 via the local network NW2 and the wide area network NW1. When the grape thinning position acquisition unit 2222 acquires the grape thinning position information indicating the grape thinning position transmitted from the cloud server 1, it notifies the display control unit 220 of the acquired grape thinning position information.

[0036] As described in Embodiment 1, the display control unit 220 forms a grape count notification image information showing a grape count notification image representing the number of grapes to be counted, based on the grape count information notified by the grape count determination unit 219, and transfers it to the display I / F 204. Furthermore, when the display control unit 220 receives grape thinning position information from the grape thinning position acquisition unit 2222, it forms a grape thinning position notification image information showing a grape thinning position notification image to notify the worker of the position of grapes that are candidates for thinning, based on the notified grape count position information, and transfers it to the display I / F 204. As a result, the grape thinning position notification image is displayed on the on-glass display unit 23. Here, both the grape count notification image and the grape thinning position notification image are displayed simultaneously on the on-glass display unit 23.

[0037] Returning to Figure 7, the cloud server 1 comprises a CPU 101, a main memory unit 102, an auxiliary memory unit 103, a communication unit 106, and a bus 109 that connects them to each other. The CPU 101 is, for example, a multi-core processor. The main memory unit 102 consists of volatile memory and is used as the work area for the CPU 101. The auxiliary memory unit 103 consists of large-capacity non-volatile memory and stores programs for realizing various functions of the cloud server 1. The communication unit 106 is connected to the wide-area network NW1.

[0038] The CPU 101 reads the program stored in the auxiliary storage unit 103 into the main storage unit 102 and executes it, so that it functions as a grape position acquisition unit 111, a grape thinning determination unit 112, a grape thinning position estimation unit 113, a model generation unit 114, and a grape thinning position notification unit 115, as shown in Figure 9. The auxiliary storage unit 103 also has a grape position storage unit 131 and a model storage unit 132. The grape position storage unit 131 stores position coordinate information indicating the relative position coordinate of each grape in the bunch image, as shown in Figure 10, for example, and associates it with grape thinning count information indicating the number of times the grapes have been thinned. In the example shown in Figure 10, after grape thinning has been performed n times, the grapes at relative position coordinates (X[0], Y[0]), (X[1], Y[1]), ..., (X[n-1], Y[n-1]) have been removed by thinning, and the grapes at relative position coordinates (X[n], Y[n]), (X[n+1], Y[n+1]), ..., (X[N], Y[N]) remain.

[0039] Returning to Figure 9, the model storage unit 132 stores information on a grape thinning position estimation model that estimates the position of grapes to be thinned based on the grape positions in the grape cluster before thinning and the positions of the thinned grapes, using the positions of grapes identified in the cluster image of the grape cluster before thinning. The grape thinning position estimation model is, for example, a recurrent neural network. In this case, the model storage unit 132 stores information indicating the structure of the grape thinning position estimation model and information indicating the weight coefficients in the grape thinning position estimation model. The information indicating the structure of the grape thinning position estimation model includes information indicating the number of nodes, the number of layers, the weight coefficients corresponding to each node, and the activation function. Here, the weight coefficients include the weight coefficients used when feeding back the output from the layer to which each node belongs to the input. This grape thinning position estimation model has an input layer L10, a hidden layer L20, and an output layer L30, as shown in Figure 11. The input layer L10 receives the relative position coordinates of each grape in the grape cluster before thinning.

[0040] The hidden layer L20 consists of three layers, each containing, for example, a predetermined number N[j] nodes y[j,i,t] (1≦i≦N[j], where N[j] is a positive integer). Here, the output y[2,j,t] of each node, excluding the node immediately following the input layer L10 and the node immediately following the output layer L30, is expressed by the following equation (1).

[0041]

number

[0042] Here, w in [k,i] and w feedback [k,j] represents the weight coefficients, and f(*) represents the activation function. Also, t represents the number of times the thinning operation has been performed. Nonlinear functions such as the sigmoid function, ramp function, and step function are used as the activation function. The output of each node at the first time point t is the output of an activation function whose arguments are the sum of the output of multiple nodes belonging to the layer preceding the node after t thinning operations have been performed, multiplied by the weight coefficients, and the sum of the output of multiple nodes belonging to the same layer as the node after the previous t-1 thinning operations have been performed, multiplied by the weight coefficients. Based on the output from the node belonging to the final layer of the hidden layer L20, the output of the output vector has as its elements the expected value for each relative position coordinate of the grape to be thinned next. Here, the output layer L30 performs processing using, for example, the softmax function. Furthermore, the model storage unit 132 also stores initial relative position coordinate information necessary when estimating the relative position coordinates of the grapes to be thinned in the bunch image using the thinning position estimation model at the start of the thinning operation.

[0043] Returning to Figure 9, when the grape position acquisition unit 111 acquires position coordinate information transmitted from the grape thinning support device 2002, it notifies the grape thinning determination unit 112 of the acquired position coordinate information. When the grape thinning determination unit 112 receives position coordinate information for a grape corresponding to a new bunch of grapes from the grape position acquisition unit 111, it stores that position coordinate information in the grape position storage unit 131. Next, when the grape thinning determination unit 112 receives position coordinate information for the same bunch of grapes from the grape position acquisition unit 111, it determines whether or not grape thinning has been performed on the grapes based on the notified position coordinate information. If the grape thinning determination unit 112 determines that grape thinning has been performed, it stores the position coordinate information notified by the grape position acquisition unit 111 in the grape position storage unit 131.

[0044] The grape thinning position estimation unit 113 estimates the relative position coordinates of the grapes to be thinned using the grape thinning position estimation model indicated by the aforementioned grape thinning position estimation model information, based on the history of position coordinate information stored in the grape position storage unit 131. Specifically, the grape thinning position estimation unit 113 calculates the expected value of each relative position coordinate of the grapes using the grape thinning position estimation model, based on the changes in the relative position coordinates of the grapes from the start of grape thinning to the most recent thinning. Then, the grape thinning position estimation unit 113 identifies a predetermined number of relative position coordinates, starting with those with the highest calculated expected values. Next, the grape thinning position estimation unit 113 notifies the grape thinning position notification unit 115 of the position coordinate information indicating the identified relative position coordinates.

[0045] The model generation unit 114 generates a grape thinning position estimation model using the history of relative position coordinates of grapes stored in the grape position storage unit 131 from the start of thinning to the most recent thinning. Specifically, the model generation unit 114 first uses the grape thinning position estimation model already stored in the model storage unit 132 to calculate the expected value of each relative position coordinate of the grapes to be thinned from the start of thinning to the thinning immediately preceding the most recent thinning. Next, the model generation unit 114 compares the relative position coordinate of the most recent grape stored in the grape position storage unit 131 with the relative position coordinate of the grape after the previous thinning and sets the expected value such that the expected value of the relative position coordinate of the actually thinned grape is higher than the expected value of the other relative position coordinates. For example, the model generation unit 114 sets the expected value of the relative position coordinate of the actually thinned grape to a number greater than 0 and sets the expected values ​​of the other relative position coordinates to "0". The model generation unit 114 then calculates the error between the expected value calculated using the grape thinning position estimation model already stored in the model storage unit 132 and the expected values ​​of the relative position coordinates of the grapes that were actually thinned and the relative position coordinates of the other grapes. The model generation unit 114 then determines the weight coefficients of the grape thinning position estimation model using the aforementioned BPTT (Back Propagation Through Time) method based on the calculated error, and updates the grape thinning position estimation model information stored in the model storage unit 132 with new grape thinning position estimation model information showing the determined weight coefficients.

[0046] The grape thinning position notification unit 115 generates grape thinning position information, which includes position coordinate information indicating the relative position coordinates of the grapes to be thinned, as notified by the grape thinning position estimation unit 113, and transmits it to the grape thinning support device 2002.

[0047] Next, the operation of the grape thinning support system according to this embodiment will be described with reference to Figure 12. It is assumed that the grape thinning support device 2002 is powered on and acquires moving image data obtained by imaging unit 22 and stores it sequentially in moving image storage unit 231. First, when a preset grape count display time arrives, the grape thinning support device 2002 selects a pair of still image data that are adjacent in time from the multiple still image data frames that constitute the moving image data stored in the moving image storage unit 231, and sequentially stores the selected pair of still image data in the selected image storage unit 229 (step S1). Here, the grape count display time is set to arrive, for example, at a period of several msec, as described in Embodiment 1. Next, each time a pair of still image data that are adjacent in time is selected and stored in the selected image storage unit 229, the grape thinning support device 2002 extracts a bunch image of at least one grape bunch from each of the pair of still image data (step S2).

[0048] Next, the grape thinning support device 2002 estimates the position of the grapes and the center position of the grape cluster in the still image data, based on the cluster image extracted from the selected pair of still image data (step S3). Subsequently, the grape thinning support device 2002 calculates the Euclidean distance between the estimated center position of the grape cluster and the center position of the still image containing the corresponding cluster image, and calculates the average value of the calculated Euclidean distances (step S4).

[0049] Next, the grape thinning support device 2002 identifies the bunch image with the smallest average value of the calculated Euclidean distance as the bunch image corresponding to the grape bunch to be counted (step S5). Subsequently, the grape thinning support device 2002 identifies bunch images corresponding to the same grape bunch based on the degree of agreement of at least one bunch image contained in each pair of still image data extracted by the bunch image extraction unit 213 (step S6). After that, the grape thinning support device 2002 generates position coordinate information corresponding to the identified bunch image (step S7), and the generated position coordinate information is transmitted from the grape thinning support device 2002 to the cloud server 1 (step S8).

[0050] Next, the grape thinning support device 2002 calculates the number of grapes in a bunch of grapes based on the bunch images corresponding to the same bunch of grapes contained in each of the extracted still image data (step S9). Subsequently, the grape thinning support device 2002 determines the average value of the number of grapes calculated for all pairs of still image data sequentially selected from multiple frames of still image data as the number of grapes in the bunch of grapes (step S10). After that, the grape thinning support device 2002 forms a grape count notification image information showing the number of grapes in the bunch of grapes to be counted based on the determined grape count information, and displays it on the onglass display unit 23 (step S11).

[0051] Furthermore, when the cloud server 1 receives notification of position coordinate information corresponding to the same bunch of grapes that has been acquired, it determines that thinning work has been performed on the grapes based on the notified position coordinate information (step S12). In this case, the cloud server 1 stores the acquired position coordinate information in the grape position storage unit 131 (step S13). Next, the cloud server 1 generates thinning position estimation model information that shows a thinning position estimation model using the history of relative position coordinates of grapes from the start of thinning to the most recent thinning, which is stored in the grape position storage unit 131, and updates the thinning position estimation model information stored in the model storage unit 132 with the generated thinning position estimation model information (step S14). Subsequently, the cloud server 1 uses the thinning position estimation model shown in the thinning position estimation model information stored in the model storage unit 132 to estimate the relative position coordinates of the grapes to be thinned from the history of position coordinate information stored in the grape position storage unit 131 (step S15). Subsequently, the cloud server 1 generates thinning position information, which includes position coordinate information indicating the relative position coordinates of the grapes to be thinned (step S16), and the generated thinning position information is transmitted from the cloud server 1 to the thinning support device 2002 (step S17).

[0052] Meanwhile, when the grape thinning support device 2002 acquires grape thinning position information, it forms a grape thinning position notification image representing the position of the grapes to be thinned based on the acquired grape thinning position information and displays it on the on-glass display unit 23 (step S18).

[0053] Furthermore, after the series of processes from steps S9 to S11 described above have been performed, the grape thinning support device 2002 determines that the number of grapes in the bunch to be counted is less than or equal to a preset standard number of grapes (step S19). In this case, the grape thinning support device 2002 outputs a notification sound to inform the operator that the number of grapes has fallen below the standard number of grapes (step S20).

[0054] Next, the grape thinning support process performed by the grape thinning support device 2002 according to this embodiment will be described with reference to Figure 13. In Figure 13, processes similar to those in Embodiment 1 are denoted by the same reference numerals as in Figure 5. First, the image selection unit 212 determines whether or not the aforementioned grape count display time has arrived (step S101). If the image selection unit 212 determines that the aforementioned grape count display time has arrived (step S101: Yes), the series of processes described in steps S102 to S107 in Embodiment 1 are executed. Next, the position coordinate notification unit 2221 generates position coordinate information corresponding to the bunch image identified by the same bunch image identification unit 2217 and transmits it to the cloud server 1 (step S2101). Subsequently, the series of processes described in steps S108 to S110 in Embodiment 1 are executed. After that, the grape count determination unit 223 determines whether or not the number of grapes determined by the grape count determination unit 219 is less than or equal to a preset standard number of grapes (step S111). If the grain count determination unit 223 determines that the number of grains determined by the grain count determination unit 219 is greater than the aforementioned reference number of grains (step S111: No), the process in step S2102 described later is executed. On the other hand, if the grain count determination unit 223 determines that the number of grains determined by the grain count determination unit 219 is less than or equal to the aforementioned reference number of grains (step S111: Yes), the notification unit 224 generates the aforementioned audio information and outputs a notification sound corresponding to the generated audio information from the speaker 26 (step S112).

[0055] Next, the grape thinning position acquisition unit 2222 determines whether or not it has acquired grape thinning position information indicating the grape thinning positions transmitted from the cloud server 1 (step S2102). If the grape thinning position acquisition unit 2222 determines that it has not acquired the grape thinning position information (step S2102: No), the process in step S101 is executed again. On the other hand, if the grape thinning position acquisition unit 2222 determines that it has acquired the grape thinning position information (step S2102: Yes), it notifies the display control unit 220 of the acquired grape thinning position information. The display control unit 220 then forms grape thinning position notification image information, which shows the grape thinning position where the operator is recommended to thin the grapes, based on the grape number position information notified by the grape thinning position acquisition unit 2222, and displays it on the onglass display unit 23 (step S2103). After that, the process in step S101 is executed again.

[0056] Next, the grape thinning position notification process executed by the cloud server 1 according to this embodiment will be explained with reference to Figure 14. This grape thinning position notification process is started, for example, when a program for executing the grape thinning position notification process is started on the cloud server 1. First, the grape position acquisition unit 111 determines whether or not it has acquired the position coordinate information transmitted from the grape thinning support device 2002 (step S201). Here, as long as the grape position acquisition unit 111 determines that it has not acquired the position coordinate information (step S201: No), it repeatedly executes the process in step S201. On the other hand, if the grape position acquisition unit 111 determines that it has acquired the position coordinate information (step S201: Yes), it notifies the grape thinning determination unit 112 of the acquired position coordinate information. Then, the grape thinning determination unit 112 determines whether or not grape thinning work has been performed on the grapes based on the notified position coordinate information (step S202). If the grape thinning determination unit 112 determines that grape thinning has not been performed (step S202: No), the process described in step S205 below is executed. On the other hand, if the grape thinning determination unit 112 determines that grape thinning has been performed (step S202: Yes), the position coordinate information notified by the grape position acquisition unit 111 is stored in the grape position storage unit 131 (step S203).

[0057] Next, the model generation unit 114 generates grape thinning position estimation model information using the history of relative position coordinates of grapes from the start of thinning to the most recent thinning, which is stored in the grape position storage unit 131. Then, the model generation unit 114 updates the grape thinning position estimation model information stored in the model storage unit 132 with the newly generated grape thinning position estimation model information (step S204).

[0058] Next, the grape thinning position estimation unit 113 estimates the relative position coordinates of the grapes to be thinned using the grape thinning position estimation model indicated by the aforementioned grape thinning position estimation model information from the history of position coordinate information stored in the grape position storage unit 131 (step S205). After that, the grape thinning position notification unit 115 generates grape thinning position information including position coordinate information indicating the relative position coordinates of the grapes to be thinned, which was estimated by the grape thinning position estimation unit 113, and transmits it to the grape thinning support device 2002 (step S206).

[0059] As described above, according to the grape thinning support system of this embodiment, the model generation unit 114 generates a grape thinning position estimation model using the history of relative position coordinates of grapes from the start of thinning to the most recent thinning. Then, the grape thinning position estimation unit 113 uses the generated grape thinning position estimation model to estimate the position of the grapes to be thinned from the position of the grapes identified in the bunch image of the grape bunch before thinning. As a result, the worker performing the thinning work can perform appropriate thinning by performing the thinning based on the estimated information on the position of the grapes to be thinned.

[0060] Although embodiments of the present invention have been described above, the present invention is not limited to the embodiments described above. For example, the notification unit 224 may change the volume of the notification sound according to the difference between the number of grapes on the bunch and the reference number of grapes described above. In this case, for example, the grape count determination unit 223 may refer to a volume table that associates the difference between the number of grapes on the bunch determined by the grape count determination unit 219 and the reference number of grapes with volume information indicating the volume of the notification sound, and notify the notification unit 224 of the volume information corresponding to the difference. The notification unit 224 may then generate audio information such that a notification sound of the volume indicated by the notified volume information is output from the speaker 26.

[0061] With this configuration, the worker can determine the number of grapes to be thinned based on the volume of the notification sound emitted from the speaker 26.

[0062] In this embodiment, for example, the notification unit 224 may change a preset notification image displayed on the on-glass display unit 23 of the glass unit 20 according to the difference between the number of grapes on the bunch and the aforementioned reference number of grapes. In this case, for example, the grape count determination unit 223 may refer to a notification image table that associates the difference between the number of grapes on the bunch determined by the grape count determination unit 219 and the reference number of grapes with notification image information indicating a preset notification image, and notify the notification unit 224 of the notification image information corresponding to the difference. The notification unit 224 may then display the notification image indicated by the notified notification image on the on-glass display unit 23.

[0063] In this embodiment, an example was described in which the number of grapes to be thinned is displayed on the on-glass display unit 23 of the glass unit 20. However, the invention is not limited to this, and information indicating the number of grapes to be thinned may be transmitted to a terminal device such as a smartphone and displayed on the display unit of the terminal device. Alternatively, the number of grapes to be thinned may not be displayed, and only an alert sound whose volume changes according to the difference between the number of grapes on the bunch of grapes to be thinned and the standard number of grapes may be output, as described above.

[0064] In this embodiment, an example was described in which the notification unit 224 outputs a notification sound. However, the invention is not limited to this, and for example, the notification unit 224 may vibrate the housing of the control unit 25 when the number of grapes on a bunch falls below a standard number.

[0065] Furthermore, the various functions of the grain thinning support device 2 and cloud server 1 according to the present invention may be realized by software, firmware, or a combination of software and firmware. In this case, the software or firmware may be written as a program, and the program may be stored on a computer-readable recording medium such as a flexible disk, CD-ROM (Compact Disc Read Only Memory), DVD (Digital Versatile Disc), and MO (Magneto-Optical Disc) and distributed, and a computer capable of realizing the aforementioned functions may be configured by loading and installing the program on a computer. In cases where each function is realized by a division of labor between the OS (Operating System) and applications, or by cooperation between the OS and applications, only the parts other than the OS may be stored on the recording medium.

[0066] Furthermore, it is possible to superimpose each program onto the carrier wave and distribute it over a network. For example, the program could be posted on a bulletin board system (BBS) on the network and distributed over the network. These programs could then be launched and executed under the control of the OS, just like other application programs, to perform the aforementioned processing. [Industrial applicability]

[0067] The present invention is suitable as a system to support grape thinning work. [Explanation of Symbols]

[0068] 1: Cloud server, 2002: Grain thinning support device, 20: Glass unit, 21: Frame, 22: Imaging unit, 23: On-glass display unit, 24: Code, 25, 2025: Control unit, 26: Speaker, 81: BBR, 82: Access point, 101, 201 CPU, 102, 202 Main memory unit, 103, 203: Auxiliary memory unit, 106: Communication unit, 109, 209: Bus, 111: Grain thinning position acquisition unit, 112: Grain thinning determination unit, 113: Grain thinning position estimation unit, 114: Model generation unit, 115: Grain thinning position notification unit, 131, 233: Grain position storage unit, 132, 232: Model storage unit, 204: Display I / F, 205: Input unit, 206: Wireless module, 207: Imaging I / F, 208: Audio I / F, 211: 212: Video acquisition unit, 213: Image selection unit, 214: Cluster image extraction unit, 215: Grain position estimation unit, 216: Distance calculation unit, 217: Counting target identification unit, 217, 2217: Same cluster image identification unit, 218: Grain number calculation unit, 219: Grain number determination unit, 220: Display control unit, 223: Grain number determination unit, 224: Notification unit, 229: Selected image storage unit, 231: Video image storage unit, 2221: Position coordinate notification unit, 2222: Grain picking position acquisition unit, NW1: Wide-area network, NW2: Local network

Claims

1. Imaging unit, A video acquisition unit that acquires moving image data obtained by moving the aforementioned imaging unit around a bunch of grapes, An image selection unit that sequentially selects a pair of still image data that are adjacent in time from the still image data of multiple frames that constitute the aforementioned moving image data, starting from the first still image data in time. Each time a pair of temporally adjacent still image data is selected, a cluster image extraction unit extracts a cluster image from each of the pair of still image data that shows at least one cluster of grapes, A unit for identifying identical grape clusters that identifies a cluster image corresponding to the same grape cluster based on the degree of similarity of at least one cluster image contained in each of the extracted pair of still image data, A grape count calculation unit calculates the number of grapes contained in each bunch image corresponding to the same bunch of grapes contained in each of the extracted still image data, The system includes a grape count determination unit that determines the average value of the number of grapes calculated for all pairs of still image data sequentially selected from multiple frames of still image data as the number of grapes on a bunch of grapes, Grain thinning support device.

2. A grain count determination unit that determines whether the determined number of grains is less than or equal to a preset standard number of grains, The system includes a notification unit that notifies the operator by voice or by displaying a notification image on the display unit when the determined number of attached particles falls below the standard number of particles. The grain thinning support device according to claim 1.

3. The notification unit notifies the worker by voice, and changes the volume of the voice as the number of attached particles approaches the standard number of particles. The grain thinning support device according to claim 2.

4. The notification unit notifies the worker with a notification image displayed on the display unit, and changes the notification image displayed on the display unit as the number of attached grains approaches the standard number of grains. The grain thinning support device according to claim 2.

5. A grape position estimation unit estimates the position of the center of the grape cluster in the still image shown by the still image data, based on the cluster image extracted from the selected pair of still image data. A distance calculation unit calculates the Euclidean distance between each identified center position of a bunch of grapes and the center position of a still image, The system further includes a counting target identification unit that identifies the bunch image with the smallest average value of the calculated Euclidean distance as the bunch image corresponding to the grape bunch for which the number of grapes is to be counted. The grain thinning support device according to any one of claims 1 to 4.

6. Imaging unit, A video acquisition unit that acquires moving image data obtained by moving the aforementioned imaging unit around a bunch of grapes, An image selection unit that sequentially selects a pair of still image data that are adjacent in time from the still image data of multiple frames that constitute the aforementioned moving image data, starting from the first still image data in time. Each time a pair of temporally adjacent still image data is selected, a cluster image extraction unit extracts a cluster image from each of the pair of still image data that shows at least one cluster of grapes, A unit for identifying identical grape clusters that identifies a cluster image corresponding to the same grape cluster based on the degree of similarity of at least one cluster image contained in each of the extracted pair of still image data, A grape count calculation unit calculates the number of grapes contained in each bunch image corresponding to the same bunch of grapes contained in each of the extracted still image data, A grape count determination unit determines the number of grapes on a bunch of grapes by calculating the average value of the number of grapes calculated for all pairs of still image data sequentially selected from multiple still image data frames, A model generation unit generates a grape thinning position estimation model using the history of the relative position coordinates of grapes from the start of thinning to the most recent thinning, The system includes a grape thinning position estimation unit that uses the aforementioned grape thinning position estimation model to estimate the position of grapes to be thinned from the position of grapes identified in a bunch image of the grape bunch before thinning, Grain thinning support system.

7. The process involves acquiring dynamic image data by moving the imaging unit around a bunch of grapes, and The steps include sequentially selecting a pair of still image data that are temporally adjacent from the still image data of multiple frames that constitute the aforementioned moving image data, starting from the first still image data in time, Each time a pair of temporally adjacent still image data is selected, the step of extracting a bunch image from each of the pair of still image data that shows at least one bunch of grapes, A step of identifying a cluster image corresponding to the same grape cluster based on the degree of similarity of at least one cluster image contained in each of the extracted pair of still image data, The steps include: calculating the number of grapes in each bunch image corresponding to the same bunch of grapes contained in each of the extracted still image data; The process includes the step of determining the average value of the number of grapes calculated for all pairs of still image data sequentially selected from multiple still image data frames as the number of grapes on the bunch, Grain thinning support method.

8. Computers, A video acquisition unit that acquires moving image data obtained by moving the imaging unit around a bunch of grapes, An image selection unit that sequentially selects a pair of still image data that are adjacent in time from the still image data of multiple frames that constitute the aforementioned moving image data, starting from the first still image data in time. Each time a pair of temporally adjacent still image data is selected, a cluster image extraction unit extracts a cluster image from each of the pair of still image data that shows at least one cluster of grapes. A cluster image identification unit identifies a cluster image corresponding to the same grape cluster based on the degree of similarity of at least one cluster image contained in each of the extracted pair of still image data. A grape count calculation unit calculates the number of grapes in each bunch image corresponding to the same bunch of grapes contained in each of the extracted still image data. A grape count determination unit determines the number of grapes on a bunch of grapes by calculating the average value of the number of grapes calculated for all pairs of still image data sequentially selected from multiple still image data frames. A program designed to function as such.

Images