Landscape analysis support apparatus, and landscape analysis support program

Abstract

To obtain a landscape analysis support apparatus and a landscape analysis support program which can contribute to making a landscape as an analysis target having better impression.SOLUTION: A landscape analysis support apparatus 10 according to the present invention has an acquiring unit 11A for acquiring image information indicating an image of a landscape as an analysis target, a deriving unit 11B for, by inputting the acquired image information into an impression evaluation model 13B, deriving impression information indicating impression of the image received by a person and influence degree information indicating, for each of predetermined regions of the image, influence degree on the impression, a detection unit 11C for detecting at least one of an element existing in at least one of a region having influence degree of a first threshold value or higher indicated by the derived influence degree information in the image and a region having the influence degree lower than the first threshold value and of a second threshold value or lower, and a feature of the element, and a presentation unit 11D for presenting the derived impression information together with at least one of the detected element and feature.SELECTED DRAWING: Figure 4

Description

Technical Field

[0001] The present invention relates to a landscape analysis support device and a landscape analysis support program.

Background Art

[0002] Conventionally, the following techniques have existed as techniques related to the analysis of landscapes for landscape images.

[0003] Patent Document 1 discloses an evaluation system including a terminal having a first display means and a reception means, and a control device. In this evaluation system, the first display means of the terminal displays a first image including an evaluation object based on first image data, and the reception means of the terminal receives, from the first image displayed on the first display means, the designation of a portion of interest by an evaluator and the input of evaluation information for the portion of interest. Further, in this evaluation system, the control device generates attention degree distribution data representing the attention degree for each position in the first image based on a plurality of received information in which information regarding the designation of the portion of interest and the evaluation information are associated, generates second image data based on the first image and the attention degree distribution data, and causes a second display means to display the second image as a second image. Then, this evaluation system determines the attention degree for each position in the first image based on information regarding a plurality of portions of interest included in the plurality of received information, and generates the attention degree distribution data according to the attention degree for each position.

[0004] Patent Document 2 discloses a landscape analysis support device aimed at enabling the basis of subjective evaluation results regarding a landscape for a region to be analyzed to be grasped.

[0005] This landscape analysis support device is a landscape analysis support device that uses a landscape analysis support model in which image information showing a captured image obtained by capturing an area to be analyzed is input information, and subjective evaluation information showing a subjective evaluation regarding the landscape of the captured image and objective evaluation information showing an objective evaluation regarding the landscape are output information. The device includes an acquisition unit that acquires the image information, a derivation unit that derives the subjective evaluation information and the objective evaluation information by inputting the image information acquired by the acquisition unit into the landscape analysis support model, and a presentation unit that presents comprehensive evaluation information which is information showing both the subjective evaluation information and the objective evaluation information derived by the derivation unit.

Prior Art Documents

Patent Documents

[0006]

Patent Document 1

Patent Document 2

Summary of the Invention

Problems to be Solved by the Invention

[0007] However, in the technologies disclosed in Patent Document 1 and Patent Document 2, although it is possible to evaluate the landscape image as a whole or partially, the impression of the landscape to be analyzed such as lively or nostalgic is not considered, and there is a problem that it is not always possible to contribute to such impression.

[0008] The present disclosure has been made in view of the above circumstances, and an object thereof is to provide a landscape analysis support device and a landscape analysis support program that can contribute to the impression of the landscape to be analyzed.

Means for Solving the Problems

[0009] The landscape analysis support device according to the present invention described in claim 1 is a landscape analysis support device using an impression evaluation model in which image information indicating an image representing a landscape to be analyzed is input information, impression information indicating an impression received by a person with respect to the image, and influence degree information indicating the degree of influence on the impression for each predetermined region of the image are output information, and includes an acquisition unit that acquires the image information, a derivation unit that derives the impression information and the influence degree information by inputting the image information acquired by the acquisition unit into the impression evaluation model, a detection unit that detects at least one of a region in the image where the degree of influence indicated by the influence degree information derived by the derivation unit is equal to or greater than a predetermined first threshold value, and a region where the degree of influence is equal to or less than a second threshold value less than the first threshold value, an element existing in at least one of the regions, and at least one of the characteristics of the element, and a presentation unit that presents the impression information derived by the derivation unit together with at least one of the element and the characteristic detected by the detection unit.

[0010] According to the landscape analysis support device according to the present invention described in claim 1, image information indicating an image representing a landscape to be analyzed is acquired, and by inputting the acquired image information into an impression evaluation model, impression information indicating an impression received by a person with respect to the image, and influence degree information indicating the degree of influence on the impression for each predetermined region of the above image are derived, at least one of a region in the above image where the degree of influence indicated by the derived influence degree information is equal to or greater than a predetermined first threshold value, and a region where the degree of influence is equal to or less than a second threshold value less than the first threshold value, an element existing in at least one of the regions, and at least one of the characteristics of the element are detected, and the derived impression information is presented together with at least one of the detected element and characteristic. Therefore, it is possible to contribute to improving the favorable impression of the landscape to be analyzed by further emphasizing at least one of the elements and characteristics that give a favorable influence and reducing at least one of the elements and characteristics that give an adverse influence.

[0011] The landscape analysis support device according to the present invention described in claim 2 is the landscape analysis support device described in claim 1, wherein the detection unit detects the element using at least one of a segmentation model using the image information as input information and an object detection model by rectangular selection.

[0012] According to the landscape analysis support device according to the present invention described in claim 2, by detecting the above element using at least one of the segmentation model using the above image information as input information and the object detection model by rectangular selection, the present invention can be implemented more simply as compared with the case where the element is detected by a person.

[0013] The landscape analysis support device according to the present invention described in claim 3 is the landscape analysis support device described in claim 1 or claim 2, further comprising an acquisition unit that further acquires instruction information indicating an instruction for correcting the above image from the user in response to the presentation by the presentation unit, and a correction unit that performs a correction according to the instruction indicated by the instruction information on the image information, wherein the derivation unit uses the image information corrected by the correction unit to derive the impression information and the influence degree information again, and the detection unit uses the image information corrected by the correction unit to detect at least one of the above element and the feature of the element again.

[0014] According to the landscape analysis support device according to the present invention described in claim 3, by further acquiring instruction information indicating an instruction for correcting the above image from the user in response to the above presentation, performing a correction according to the instruction indicated by the instruction information on the image information, using the corrected image information to derive the impression information and the influence degree information again, and using the corrected image information to detect at least one of the above element and the feature of the element again, it is possible to efficiently improve the favorable impression of the landscape on the analysis target.

[0015] The landscape analysis support program according to the present invention described in claim 4 is a landscape analysis support program using an impression evaluation model in which image information indicating an image representing a landscape to be analyzed is input information, impression information indicating an impression received by a person with respect to the image, and influence degree information indicating the degree of influence on the impression for each predetermined region of the image are output information. The program acquires the image information, inputs the acquired image information into the impression evaluation model to derive the impression information and the influence degree information, detects at least one of an element existing in at least one of a region in the image where the influence degree indicated by the derived influence degree information is equal to or greater than a predetermined first threshold value and a region where the influence degree is equal to or less than a second threshold value less than the first threshold value, and at least one of the characteristics of the element, and causes a computer to execute a process of presenting the derived impression information together with at least one of the detected element and the characteristic.

[0016] According to the landscape analysis support program according to the present invention described in claim 4, by acquiring image information indicating an image representing a landscape to be analyzed and inputting the acquired image information into an impression evaluation model, impression information indicating an impression received by a person with respect to the image, and influence degree information indicating the degree of influence on the impression for each predetermined region of the above image are derived. At least one of an element existing in at least one of a region in the above image where the influence degree indicated by the derived influence degree information is equal to or greater than a predetermined first threshold value and a region where the influence degree is equal to or less than a second threshold value less than the first threshold value, and at least one of the characteristics of the element are detected, and the derived impression information is presented together with at least one of the detected element and the characteristic. Therefore, it is possible to contribute to the favorable impression of the landscape to be analyzed by further emphasizing at least one of the elements and characteristics that give a favorable influence and reducing at least one of the elements and characteristics that give an unfavorable influence.

Effects of the Invention

[0017] As described above, according to the present invention, it is possible to contribute to the favorable impression of the landscape to be analyzed.

Brief Description of the Drawings

[0018]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Embodiments for Carrying Out the Invention

[0019] Hereinafter, with reference to the drawings, an example of an embodiment for carrying out the present invention will be described in detail.

[0020] First, with reference to FIG. 1, the configuration of the landscape analysis support system 1 according to the present embodiment will be described. FIG. 1 is a block diagram showing an example of the configuration of the landscape analysis support system 1 according to the present embodiment.

[0021] As shown in FIG. 1, the landscape analysis support system 1 according to the present embodiment includes a landscape analysis support device 10 that plays a central role in this system, and an information storage device 90. The landscape analysis support device 10 according to the present embodiment uses an impression evaluation model and an element detection model that are constructed in advance as machine learning models to present information regarding the impression received by a person with respect to an image representing a landscape to be analyzed, and is intended to contribute to the favorable impression of the landscape thereby. Further, the information storage device 90 according to the present embodiment stores various information handled by the landscape analysis support device 10.

[0022] The information storage device 90 according to the present embodiment includes a non-volatile storage unit 92. The storage unit 92 is realized by an HDD (Hard Disk Drive), an SSD (Solid State Drive), a flash memory, or the like. An image information database 92A is stored in the storage unit 92 as a storage medium. Details of the image information database 92A will be described later. In addition to the image information database 92A, information used for learning the impression evaluation model and the element detection model, the details of which will be described later, is stored in the storage unit 92, but the description of these data is omitted.

[0023] The landscape analysis support device 10 and the information storage device 90 are connected via a network N, and the landscape analysis support device 10 can communicate with each other via the information storage device 90 and the network N. In the present embodiment, an in-company communication line such as a LAN (Local Area Network) or a WAN (Wide Area Network) is applied as the network N, but it is not limited to this form. For example, a public communication line such as the Internet or a telephone line may be applied as the network N, or a combination of these in-company communication lines and public communication lines may be applied. Further, in the present embodiment, a wired communication line is applied as the network N, but it is not limited to this form, and a wireless communication line may be applied, or a combination of wired and wireless communication lines may be applied.

[0024] Next, referring to FIG. 2, the hardware configuration of the landscape analysis support apparatus 10 according to the present embodiment will be described. FIG. 2 is a block diagram showing an example of the hardware configuration of the landscape analysis support apparatus 10 according to the present embodiment. Note that examples of the landscape analysis support apparatus 10 include various computers such as personal computers and server computers.

[0025] The landscape analysis support apparatus 10 according to the present embodiment includes a CPU (Central Processing Unit) 11, a memory 12 as a temporary storage area, a nonvolatile storage unit 13, an input unit 14 such as a keyboard and a mouse, a display unit 15 such as a liquid crystal display, a medium reading / writing device (R / W) 16, and a communication interface (I / F) unit 18. The CPU 11, the memory 12, the storage unit 13, the input unit 14, the display unit 15, the medium reading / writing device 16, and the communication I / F unit 18 are connected to each other via a bus B. The medium reading / writing device 16 reads information written on a recording medium 17 and writes information to the recording medium 17.

[0026] The storage unit 13 is realized by an HDD, an SSD, a flash memory, or the like. A landscape analysis support program 13A is stored in the storage unit 13 as a storage medium. The landscape analysis support program 13A is stored (installed) in the storage unit 13 by setting the recording medium 17 on which the program 13A is written in the medium reading / writing device 16 and reading the program 13A from the recording medium 17 by the medium reading / writing device 16. The CPU 11 appropriately reads the landscape analysis support program 13A from the storage unit 13, expands it in the memory 12, and sequentially executes processes included in the program 13A.

[0027] In addition, an impression evaluation model 13B is stored in the storage unit 13. The impression evaluation model 13B according to the present embodiment has input information as image information showing an image (hereinafter referred to as a "landscape image") representing a landscape to be analyzed, and output information as impression information showing an impression that a person receives from the landscape image and influence degree information showing the degree of influence on the impression for each predetermined area of the landscape image.

[0028] The impression evaluation model 13B according to this embodiment is a model based on an AI (Artificial Intelligence) using a CNN (Convolutional Neural Network), which is machine-learned in advance with image information showing landscape images (photo images in this embodiment) acquired in the past as input information and impression information showing the impressions received by people with respect to the landscape images as output information, and applies a method of visualizing the region serving as the basis for the output impression information as a heatmap.

[0029] In this embodiment, Grad (Gradient-weighted)-CAM (Class Activation Mapping) is applied as the visualization method, but it is not limited thereto. For example, Grad-CAM++, Score-CAM, Group-CAM, LIME (Local Interpretable Model-agnostic Explanations), etc. may be applied as the visualization method. Also, it may be applied based on other machine learning models such as AI other than the above CNN.

[0030] Also, as described above, in this embodiment, the case where a model applying a visualization method to a machine learning model by CNN is applied as the impression evaluation model 13B will be described, but it is not limited thereto. For example, by applying the above influence degree information as output information in addition to the impression information to the above-described machine learning model by CNN, a machine learning model to which the above visualization method is not applied may be applied as the impression evaluation model 13B.

[0031] In the landscape analysis support system 1 according to the present embodiment, as the types of impressions, both types related to favorable impressions usually required for landscapes, such as "peaceful", "high quality", "lively", "nostalgic", "healing", etc., and types related to unfavorable impressions usually not required for landscapes, such as "dreary", "unstable", "lacking in personality", etc., are applied, but it is not limited to this. For example, it may be in a form where only one of the types related to the favorable impressions and the types related to the unfavorable impressions is applied as the types of impressions.

[0032] Furthermore, an element detection model 13C is stored in the storage unit 13. In the landscape analysis support system 1 according to the present embodiment, as the element detection model 13C, a model using the method of semantic segmentation using deep learning technology is applied. FIG. 3 shows examples of images before and after classification for each element by the image segmentation technology. Note that the left image in FIG. 3 is an example of an image before classification (denoted as "original image" in FIG. 3), and the right image in FIG. 3 is an example of an image after classification for the said image (denoted as "segmentation image" in FIG. 3). Also, in the example shown in FIG. 3, vegetation, sidewalk, roadway, building, and sky are applied as constituent elements, and each constituent element is represented with a different density.

[0033] Also, in the present embodiment, teacher data using photographed images showing landscapes in various places acquired in the past is created to train the element detection model 13C. However, it is not limited to this form, and it may also be in a form where the element detection model 13C is trained using a parse created using 3D CAD (Computer Aided Design) as a landscape image.

[0034] Note that the technology of semantic segmentation using deep learning technology is also described in "ICUC10_Fang-Ying GONG_GSV-ViewFactorMaps_6-10 August 2018, 'Quantification of street elements in hyper-dense urban environments in climate research using Google Street View'", "Badrinarayanan, V., Kendall, A., & Cipolla, R. (2017). Segnet: A deep convolutional encoder-decoder architecture for image segmentation. IEEE transactions on pattern analysis and machine intelligence, 39(12), 2481-2495.", "Chen, L. C., Papandreou, G., Kokkinos, I., Murphy, K., & Yuille, A. L. (2017). Deeplab: Semantic image segmentation with deep convolutional nets, atrous convolution, and fully connected crfs. IEEE transactions on pattern analysis and machine intelligence, 40(4), 834-848.", etc. Since it is a widely known technology, further explanation here is omitted.

[0035] Thus, in the landscape analysis support system 1 according to this embodiment, as the element detection model 13C, a model using the semantic segmentation method using deep learning technology is applied, but it is not limited to this form. For example, a model applying other image segmentation technologies such as instance-aware segmentation may be applied as the element detection model 13C.

[0036] Next, with reference to FIG. 4, the functional configuration of the landscape analysis support apparatus 10 according to the present embodiment during the operation of the impression evaluation model 13B and the element detection model 13C will be described. FIG. 4 is a block diagram showing an example of the functional configuration of the landscape analysis support apparatus 10 according to the present embodiment during the operation of the impression evaluation model 13B and the element detection model 13C.

[0037] As shown in FIG. 4, the landscape analysis support apparatus 10 during the operation of the impression evaluation model 13B and the element detection model 13C includes an acquisition unit 11A, a derivation unit 11B, a detection unit 11C, a presentation unit 11D, and a correction unit 11E. When the CPU 11 of the landscape analysis support apparatus 10 executes the landscape analysis support program 13A, it functions as the acquisition unit 11A, the derivation unit 11B, the detection unit 11C, the presentation unit 11D, and the correction unit 11E.

[0038] The acquisition unit 11A according to the present embodiment acquires image information (hereinafter simply referred to as "image information") indicating a landscape image to be analyzed.

[0039] In the present embodiment, the above image information is acquired by reading it from an image information database 92A (see also FIG. 5) described later. However, it is not limited to this form. For example, one of the image information regarding an unspecified number of landscapes posted on an SNS (Social Networking Service) or the like may be directly acquired as the image information via the communication I / F unit 18 or the like.

[0040] Further, the derivation unit 11B according to the present embodiment inputs the image information acquired by the acquisition unit 11A into the impression evaluation model 13B to derive the above-described impression information and influence degree information.

[0041] On the one hand, the detection unit 11C according to the present embodiment detects at least one of the regions in the landscape image where the degree of influence indicated by the degree of influence information derived by the derivation unit 11B is equal to or greater than a predetermined first threshold value, and the regions where the degree of influence is equal to or less than a second threshold value less than the first threshold value (in the present embodiment, both regions), and at least one of the features of the element (in the present embodiment, only the element). In the present embodiment, as the first threshold value, a value of 30% of the numerical range that the degree of influence can take is applied, and as the second threshold value, a value of 5% in the numerical range is applied. However, it goes without saying that none of the threshold values are limited to these values. Also, although not applied in the present embodiment, the features of the element include the sky ratio when the element is empty, the road ratio when the element is a road, the green view ratio when the element is a planting, and the like.

[0042] Then, the presentation unit 11D according to the present embodiment presents the impression information derived by the derivation unit 11B together with at least one of the element and the feature (in the present embodiment, the element) detected by the detection unit 11C. In the present embodiment, as the presentation by the presentation unit 11D, presentation by display on the display unit 15 is applied, but it is not limited thereto. For example, a form in which presentation by voice by an audio playback device, presentation by printing by an image forming device, or the like is applied as the presentation by the presentation unit 11D may also be used.

[0043] Here, the detection unit 11C according to the present embodiment detects the above element using the element detection model 13C that is the above-described segmentation model, but is not limited thereto. For example, the detection unit 11C may detect the above element using an object detection model by a conventionally known rectangular selection, or may apply a combination of these techniques.

[0044] Furthermore, the acquisition unit 11A according to the present embodiment further acquires instruction information indicating an instruction for correcting the landscape image from the user in response to the presentation by the presentation unit 11D. The correction unit 11E according to the present embodiment performs correction on the image information according to the instruction indicated by the instruction information. Then, the derivation unit 11B according to the present embodiment re-derives the impression information and the influence degree information using the image information corrected by the correction unit 11E. On the other hand, the detection unit 11C according to the present embodiment re-detects at least one of the above elements and features (in this embodiment, the element) using the image information corrected by the correction unit 11E.

[0045] By re-performing the derivation of the impression information and the influence degree information by the derivation unit 11B and the detection of the element by the detection unit 11C, the display content on the display unit 15 by the presentation unit 11D can be updated to use the corrected image information. As a result, the favorable impression of the landscape for the analysis target can be efficiently achieved.

[0046] Next, with reference to FIG. 5, the image information database 92A according to the present embodiment will be described. FIG. 5 is a schematic diagram showing an example of the configuration of the image information database 92A according to the present embodiment. The image information database 92A according to the present embodiment is for storing image information (hereinafter referred to as "analysis target image information") indicating a landscape image to be analyzed.

[0047] As shown in FIG. 5, in the image information database 92A according to the present embodiment, each piece of information including the region name, the analysis target image ID (Identification), and the analysis target image information is stored in an associated manner.

[0048] The above region name is information indicating the name of the region where shooting was performed to obtain the corresponding analysis target image information. The above analysis target image ID is information pre-assigned as different information for each of the analysis target image information in order to individually identify the corresponding analysis target image information. The above analysis target image information is information indicating the analysis target image information itself.

[0049] Next, with reference to FIGS. 6 to 8, the operation of the landscape analysis support device 10 when the impression evaluation model 13B and the element detection model 13C are applied will be described. FIG. 6 is a flowchart showing an example of the landscape analysis support process according to the present embodiment.

[0050] When the CPU 11 of the landscape analysis support device 10 executes the landscape analysis support program 13A, the landscape analysis support process shown in FIG. 6 is executed. The landscape analysis support process shown in FIG. 6 is executed when an instruction input to start the execution of the landscape analysis support program 13A is input by the user via the input unit 14. In order to avoid complication, hereinafter, the case where the learning of the impression evaluation model 13B and the element detection model 13C has been completed and the image information database 92A has already been constructed will be described.

[0051] In step 100 of FIG. 6, the CPU 11 controls the display unit 15 to display an initial information input screen having a predetermined configuration. In step 102, the CPU 11 waits until predetermined information is input.

[0052] FIG. 7 shows an example of the configuration of the initial information input screen according to the present embodiment. As shown in FIG. 7, on the initial information input screen according to the present embodiment, a message prompting the input of the name of the target area to be analyzed for the landscape (hereinafter referred to as the "target area name") is displayed. Further, on the initial information input screen according to the present embodiment, an input area 15A for inputting the target area name is displayed.

[0053] As an example, when the initial information input screen shown in FIG. 7 is displayed on the display unit 15, the user uses the input unit 14 to input the target area name to be analyzed for the landscape into the input area 15A and then designates the end button 15G.

[0054] When the end button 15G is designated by the user, step 102 becomes an affirmative determination and the process proceeds to step 104.

[0055] In step 104, the CPU 11 reads out, from the image information database 92A, the analysis image information corresponding to the target area indicated by the input target area name. In step 106, the CPU 11 inputs the read analysis image information into the impression evaluation model 13B. In step 108, the CPU 11 acquires each piece of information, i.e., the impression information and the influence degree information, output from the impression evaluation model 13B in response to the input of the analysis image information.

[0056] In step 110, the CPU 11 inputs the read analysis image information into the element detection model 13C. In step 112, the CPU 11 acquires the element information indicating the elements output from the element detection model 13C in response to the input of the analysis image information. Since the element detection model 13C according to the present embodiment is configured by the segmentation model as described above, as an example, as shown in the right diagram of FIG. 3, for each predetermined division area (in this embodiment, for each pixel) in the landscape image indicated by the analysis image information, information indicating the type of the element can be obtained as the element information.

[0057] Therefore, in step 114, the CPU 11 controls the display unit 15 to display an evaluation result screen having a predetermined configuration, using the impression information, the influence degree information, and the element information obtained by the above processing. In step 116, the CPU 11 waits until predetermined information is input.

[0058] FIG. 8 shows an example of the configuration of the evaluation result screen according to the present embodiment. As shown in FIG. 8, in the evaluation result screen according to the present embodiment, the evaluation target image 15C1 indicated by the analysis image information is displayed, and the name of the impression indicated by the impression information obtained by the impression evaluation model 13B (in the example shown in FIG. 8, the name "There is a sense of calmness") is displayed. Further, in the evaluation result screen according to the present embodiment, the influence degree on the impression of the name is equal to or higher than the first threshold value, and the name of the largest element (in the example shown in FIG. 8, "roof", hereinafter referred to as the "maximum influence element"), and the influence degree is equal to or lower than the second threshold value, and the name of the smallest element (in the example shown in FIG. 8, "road surface", hereinafter referred to as the "minimum influence element") are displayed.

[0059] In addition, on the evaluation result screen according to the present embodiment, an overlapping image 15C2 in which a heat map indicating influence degree information is superimposed on the evaluation target image 15C1 is enlarged and displayed, and an image 15C3 of a region corresponding to the maximum influence element in the overlapping image 15C2 and an image 15C4 of a region corresponding to the minimum influence element in the overlapping image 15C2 are enlarged and displayed together with the names of the corresponding elements and the influence degrees.

[0060] Therefore, by referring to the evaluation result screen, the user can intuitively grasp the impression that a person is likely to receive for the evaluation target image 15C1, together with the elements serving as the basis and the degree of influence of the elements on the impression.

[0061] As an example, when the evaluation result screen shown in FIG. 8 is displayed on the display unit 15, the user grasps the evaluation content. Then, when the user wants to correct the evaluation target image 15C1, the user designates the correction button 15F using the input unit 14, while when the user wants to end the display of the evaluation result screen, the user designates the end button 15G using the input unit 14. When the correction button 15F or the end button 15G is designated by the user, step 116 becomes an affirmative determination and the process proceeds to step 118.

[0062] In step 118, the CPU 11 determines whether the correction button 15F has been designated by the user. If the determination is affirmative, the process proceeds to step 120, where after executing a correction process for correcting the evaluation target image 15C1, the process returns to step 106. In the present embodiment, as the above correction process, correction using general-purpose photo retouch software is applied, but it goes without saying that it is not limited to this. Information indicating an instruction for correcting the image input from the user in this correction process corresponds to the above-described instruction information.

[0063] On the other hand, if the determination in the process of step 118 is negative, it is considered that the end button 15G has been designated by the user, and the present landscape analysis support process is ended.

[0064] As described above, the landscape analysis support device 10 according to the present embodiment acquires image information indicating an image representing a landscape to be analyzed, and inputs the acquired image information into an impression evaluation model, thereby obtaining impression information indicating the impression received by a person for the image, and influence degree information indicating the degree of influence on the impression for each predetermined region of the image. The device detects at least one of: a region in the image where the degree of influence indicated by the derived influence degree information is equal to or greater than a predetermined first threshold value; and a region where the degree of influence is equal to or less than a second threshold value that is less than the first threshold value, an element existing in at least one of these regions, and at least one of the characteristics of the element, and presents the derived impression information together with at least one of the detected element and characteristics. Therefore, by further emphasizing at least one of the elements and characteristics that give a favorable influence and reducing at least one of the elements and characteristics that give an unfavorable influence, it is possible to contribute to improving the favorable impression of the landscape to be analyzed.

[0065] Further, according to the present embodiment, the element is detected by using at least one of a segmentation model using the image information as input information and an object detection model by rectangular selection. Therefore, the present invention can be implemented more simply as compared with the case where the element is detected by a person.

[0066] Furthermore, according to the present embodiment, instruction information indicating a modification instruction for the image from the user corresponding to the presentation is further acquired, a modification corresponding to the instruction indicated by the instruction information is performed on the image information, the derivation of the impression information and the influence degree information is performed again using the modified image information, and the detection of at least one of the element and the characteristics of the element is performed again using the modified image information. Therefore, it is possible to efficiently improve the favorable impression of the landscape for the analysis target.

[0067] Note that the configuration of the evaluation result screen illustrated in the above embodiment is merely an example, and it goes without saying that modifications may be made. For example, only one of the images 15C3 and 15C4 shown in FIG. 8 may be displayed, or the evaluation target image 15C1 may be displayed in the same size as the superimposed image 15C2 without enlarging the superimposed image 15C2.

[0068] In the above-described embodiment, the case where a photographic image is applied as an image showing the landscape to be analyzed has been described, but the present invention is not limited to this. For example, a form in which a parse created using 3D CAD such as Rhinoceros (registered trademark) is applied as an image showing the landscape to be analyzed may be used. In this case, in the correction process of step 120 in the landscape analysis support process, the above parse will be corrected using the above 3D CAD.

[0069] Also, in the above-described embodiment, although not particularly mentioned, various information obtained by the landscape analysis support process may be stored by being database-ized or the like, and not only displayed, but also reused for other evaluations and the like.

[0070] In the above-described embodiment, the case where the landscape analysis support device of the present invention is applied as the landscape analysis support system 1 using the landscape analysis support device 10 and the information storage device 90 has been described, but the present invention is not limited to this. For example, a form in which the landscape analysis support device of the present invention is configured as a single device in which the landscape analysis support device 10 and the information storage device 90 are integrated may be used.

[0071] Also, in the above-described embodiment, for example, as the hardware structure of a processing unit that executes each process of the acquisition unit 11A, the derivation unit 11B, the detection unit 11C, the presentation unit 11D, and the correction unit 11E, various processors shown below can be used. As described above, the above various processors include, in addition to a CPU, which is a general-purpose processor that executes software (program) and functions as a processing unit, a programmable logic device (PLD), which is a processor whose circuit configuration can be changed after manufacture, such as an FPGA (Field-Programmable Gate Array), and a dedicated electric circuit, which is a processor having a circuit configuration designed specifically to execute specific processing, such as an ASIC (Application Specific Integrated Circuit).

[0072] The processing unit may be composed of one of these various processors, or may be composed of a combination of two or more processors of the same type or different types (for example, a combination of multiple FPGAs, or a combination of a CPU and an FPGA). Also, the processing unit may be composed of one processor.

[0073] As an example of configuring the processing unit with one processor, firstly, as represented by computers such as clients and servers, there is a form in which one processor is configured by a combination of one or more CPUs and software, and this processor functions as the processing unit. Secondly, as represented by a System On Chip (SoC), etc., there is a form in which a processor that realizes the functions of the entire system including the processing unit with one IC (Integrated Circuit) chip is used. Thus, the processing unit is configured using one or more of the above various processors as a hardware structure.

[0074] Furthermore, as a hardware structure of these various processors, more specifically, an electrical circuit (circuitry) combined with circuit elements such as semiconductor elements can be used.

Explanation of Signs

[0075] 1 Landscape analysis support system 10 Landscape analysis support device 11 CPU 11A Acquisition unit 11B Derivation unit 11C Detection unit 11D Presentation unit 11E Correction unit 12 Memory 13 Storage unit 13A Landscape analysis support program 13B Impression evaluation model 13C Element detection model 14 Input unit 15 Display unit 15A Input area 15C1 Image to be evaluated 15C2 Overlay image 15C3 Image 15C4 Image 15F Correction button 15G End button 16 Media reading / writing device 17 Recording medium 18 Communication I / F section 90 Information storage device 92 Memory section 92A Image information database N Network

Claims

1. A landscape analysis support device using an impression evaluation model in which image information showing an image representing a landscape to be analyzed is input information, impression information indicating an impression received by a person with respect to the image, and influence degree information indicating the degree of influence on the impression for each predetermined region of the image are output information, comprising: an acquisition unit that acquires the image information; a derivation unit that derives the impression information and the influence degree information by inputting the image information acquired by the acquisition unit into the impression evaluation model; a detection unit that detects at least one of an element existing in at least one of a region in the image where the degree of influence indicated by the influence degree information derived by the derivation unit is equal to or greater than a predetermined first threshold value and a region where the degree of influence is equal to or less than a second threshold value less than the first threshold value, and at least one of the characteristics of the element; a presentation unit that presents the impression information derived by the derivation unit together with at least one of the element and the characteristics detected by the detection unit; A landscape analysis support device comprising the above components.

2. The detection unit detects the element using at least one of a segmentation model using the image information as input information and an object detection model by rectangular selection. The landscape analysis support device according to Claim 1.

3. The acquisition unit further acquires instruction information indicating an instruction for correcting the image from the user in response to the presentation by the presentation unit, further comprising a correction unit that performs a correction on the image information according to the instruction indicated by the instruction information, the derivation unit performs derivation of the impression information and the influence degree information again using the image information corrected by the correction unit, the detection unit performs detection of at least one of the element and the characteristics of the element again using the image information corrected by the correction unit. The landscape analysis support device according to Claim 1 or Claim 2.

4. A landscape analysis support program using an impression evaluation model in which image information showing an image representing a landscape to be analyzed is input information, impression information indicating an impression received by a person with respect to the image, and influence degree information indicating the degree of influence on the impression for each predetermined region of the image are output information, comprising: acquiring the image information; deriving the impression information and the influence degree information by inputting the acquired image information into the impression evaluation model; An element that exists in at least one of a region in the image where the degree of influence indicated by the derived influence degree information is equal to or greater than a first threshold value determined in advance, and a region where the degree of influence is equal to or less than a second threshold value less than the first threshold value, and at least one of the characteristics of the element are detected. The derived impression information is presented together with at least one of the detected element and the characteristic. A landscape analysis support program that causes a computer to execute the process.

Images