Image analysis system, image analysis program, and image analysis method

Abstract

To provide an image analysis system capable of displaying meaning of a numerical analysis result generated for each design element of a design image and an improvement policy for the design image in an easily comprehensible manner.SOLUTION: An image analysis system 1 provided herein comprises a control unit 11 configured to obtain an analysis result for each design element of each of multiple design images and compute statistical values using the analysis results. The image analysis system 1 also comprises a display unit 16 configured to display images related to at least one of the design images together with an analysis result and statistical value for each design element of the at least one design image.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to an image analysis system, an image analysis program, and an image analysis method. [Background technology]

[0002] BACKGROUND ART Conventionally, as an information processing device for performing image analysis, a device is known that generates difference information between features of a plurality of package designs and evaluates the conspicuousness of the package based on this difference information (see Patent Document 1). Also known is a method for generating a statistical model in which the result of the sensory evaluation for each predetermined evaluation item is used as an explained variable and the layout feature amount is used as an explanatory variable (see Patent Document 2). In such a method, the result of the sensory evaluation is derived based on the statistical model and the layout feature amount. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Patent No. 7279815 [Patent Document 2] Japanese Patent Application Laid-Open No. 2006-85523 Summary of the Invention [Problem to be solved by the invention]

[0004] However, in the invention described in Patent Document 1, difference information of design images is generated using a machine learning model, which makes it difficult to quantitatively identify image features that contribute to differences for each design element that can be understood by a design creator. Furthermore, in the invention described in Patent Document 2, it is necessary to obtain in advance the affective evaluation value that serves as the explained variable according to the product category in which the design is utilized. For this reason, the user has to spend a lot of time and effort acquiring the necessary affective evaluation value data and inputting the data in a categorized manner, making the analysis process complicated. Furthermore, the meaning of the numerical values of the analysis results output for each design element is difficult to understand. Also, useful information such as design improvement policies cannot be obtained, so further improvement is required.

[0005] The image analysis system of the present invention has been developed in light of this background. That is, it aims to clearly display the meaning of the numerical values of the analysis results output for each design element of a design image and a policy for improving the design image. [Means for solving the problem]

[0006] The above-mentioned problems of the present invention can be solved by the following configuration. (1) An image analysis system includes a control unit that acquires analysis results for each design element of a plurality of design images and calculates statistical values using the analysis results. The image analysis system also includes a display unit that displays the analysis results and statistical values for at least one of the design images.

[0007] (2) The image analysis system described in (1), wherein the statistical value includes any of deviation values, standard deviations, mean values, and medians calculated using the analysis results of analyzing multiple design images for each design element based on statistical theory.

[0008] (3) The image analysis system described in (1) or (2), wherein the display unit displays a list of multiple design images from which the analysis results were obtained, and the analysis results and statistical values analyzed from the design elements of each design image.

[0009] (4) The image analysis system described in (3), wherein the control unit causes the display unit to display explanatory information for the user to interpret the list display displayed on the display unit.

[0010] (5) The image analysis system described in (4), wherein the explanatory information is information regarding the meaning of the displayed numerical value, or any of the psychological effects, emotions, or marketing effects that may occur to the observer of the image when the image has the characteristics indicated by the numerical value.

[0011] (6) The image analysis system described in (1), wherein the display unit displays the importance of the design elements qualitatively or quantitatively depending on the medium to which the design image that the user is evaluating is applied.

[0012] (7) The image analysis system described in (1), wherein the display unit displays variables with high explanatory power from variables contained in image-related data or variables contained in related data other than images for data analysis results using at least one of analytical processing data, other data extracted from multiple design images, or data integrating these.

[0013] (8) The image analysis system according to (4), wherein the display unit displays an area where the difference between the evaluation target image and the comparison image is within a predetermined value range using a graph image.

[0014] (9) The image analysis system described in (1), wherein the control unit, when analyzing the design elements of a design image, generates the similar design image using artificial intelligence and adds the generated similar design image to analyze the design elements.

[0015] (10) The image analysis program causes a computer to execute the steps of: obtaining analysis results for each design element of a plurality of design images; calculating statistical values using the analysis results; and displaying the analysis results and statistical values of at least one of the design images together with the design image.

[0016] (11) The image analysis method includes the steps of obtaining analysis results for each design element of a plurality of design images, calculating statistical values using the analysis results, and displaying the analysis results and statistical values for at least one of the design images together with the design image. [Effects of the Invention]

[0017] According to the present invention, it is possible to display in an easy-to-understand manner the meaning of the numerical values of the analysis results output for each design element of a design image, and a policy for improving the design image. [Brief explanation of the drawings]

[0018] [Figure 1] 1 is a block diagram showing a configuration of an image analysis system according to an embodiment of the present invention. [Figure 2] 10 is a screen in the first embodiment in which analysis results and statistical values are output and displayed together with a design image. [Figure 3] 10 is a screen in the second embodiment in which analysis results and statistical values are visualized as graphs and output and displayed. [Figure 4] 10 is a flowchart showing an analysis process according to a second embodiment. [Figure 5] 1 is a principal component analysis graph. DETAILED DESCRIPTION OF THE INVENTION

[0019] [Image analysis system configuration] DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings. FIG. 1 is a configuration diagram of an image analysis system 1 according to this embodiment. The image analysis system 1 analyzes multiple images input by the user and displays, for each design element, absolute evaluation values that quantify these elements and the deviation values of these absolute evaluation values as relative evaluation values. The image analysis system 1 also changes the importance of evaluation items for each medium to which the design image is applied.

[0020] Image analysis system 1 mainly comprises a control unit 11, a memory unit 12, a communication interface unit 13, an operation input unit 14, a data input unit 15, and a display unit 16. Control unit 11, memory unit 12, communication interface unit 13, operation input unit 14, data input unit 15, and display unit 16 are connected to each other via a bus 18.

[0021] Of these, the control unit 11 includes a CPU (Central Processing Unit). The CPU reads a program corresponding to the processing content from a ROM (Read Only Memory) 12a or the like included in the storage unit 12, and loads the program into a RAM (Random Access Memory) 12b. The storage unit 12 further stores an image database 12c and an image analysis program 12d. Then, the CPU of the control unit 11 reads and executes the image analysis program 12d, processes the design image data stored in the image database (described as "Image DB (Data Base)" in Figure 1) of the memory unit 12, and displays the processing results on the display unit 16.

[0022] In the image analysis system 1 of this embodiment, a design image is associated with at least one image related to the design image. The image related to the design image may be image data obtained by processing the design image, for example, by reducing the size or resolution to the extent that the corresponding design image can be identified. In this embodiment, the design images prepared are an evaluation target image created by the user and existing comparative images #1 to #3 that are already on the market (see FIG. 2). Comparative images #1 to #3 are examples, and it is preferable to prepare a statistically significant number of comparative images (reference image group).

[0023] The control unit 11 generates output data from the image data of the package design image and various related data to be output and displayed on the display unit 16. The generated output data is labeled (associated) with the package design image, and is output to the display unit 16 as is and stored in the storage unit 12. Then, the control unit 11 acquires an absolute evaluation value, which is the analysis result of each design element, for each of the plurality of design images. Furthermore, the control unit 11 calculates statistical values using the absolute evaluation values, which are the analysis results. The statistical values include values calculated based on statistical theory, such as deviation values, standard deviations, average values, and medians, which are calculated using the absolute evaluation values obtained by analyzing the plurality of design images for each design element based on statistical theory.

[0024] The storage unit 12 can store a plurality of design images, distinguishing between data of the user's design images, which are images to be evaluated, and data of a reference image group. Here, the reference image group is, for example, design images of competitors collected by the user, and is also used as comparison images. In addition, the memory unit 12 can label (associate) each design image with the absolute evaluation value, which is the analysis result of each design element of each design image, and the statistical value calculated from the absolute evaluation value of each design image, and store them as output data corresponding to each design image.

[0025] The storage unit 12 can also store at least one of preset evaluation items, such as the attention level of a specified location, blank space ratio, area ratio of a character area, GLCM (contrast), diagonal component, average brightness, average saturation, color difference ΔE00, and base color area ratio. As evaluation items in the embodiment, design elements that are well known as analysis items for design images are used. Furthermore, evaluation items other than the evaluation items such as the attention level of these designated portions may be stored in the storage unit 12.

[0026] The communication interface unit 13 controls the transmission and reception of data between the image analysis system 1 and an external device in accordance with a predetermined communication standard. As a result, a part of the image analysis system 1, such as the storage unit 12, may be distributed and arranged on a network such as a cloud connected via the communication interface unit 13.

[0027] The operation input unit 14 has a keyboard, a mouse, various pointing devices, etc., and in addition to or instead of these, a touch panel, etc., positioned so as to overlap the display screen of the display unit 16. Then, by operating the operation input unit 14, the control unit 11 calls up output data labeled with the desired design image from the storage unit 12, and outputs and displays it on the display unit 16.

[0028] The data input unit 15 has a port that allows image data of a design image to be input by inserting a storage such as an SSD (Solid State Drive) or a memory card via a USB (Universal Serial Bus: registered trademark) interface. This allows the data input unit 15 to acquire (input into the image analysis system 1) image data (hereinafter referred to as "evaluation target image") that will be the subject of saliency analysis (evaluation target) from an external device, etc. (for example, a dedicated terminal used to create designs, etc.). The communication interface unit 13 also receives and acquires image data from an external device on a wired or wirelessly connected network. The data input unit 15 may be configured to receive and acquire image data from the communication interface unit 13 in addition to or instead of the data input unit 15.

[0029] The display unit 16 has a display screen configured by a monitor such as a liquid crystal display (LCD). A touch panel that functions as the operation input unit 14 may be integrally formed on the display screen.

[0030] Then, when the user operates the operation input unit 14, the control unit 11 outputs output data labeled with the desired design image or output data called from the memory unit 12 onto the display screen of the display unit 16.

[0031] As a result, various images (e.g., evaluation target images, saliency analysis results, etc.) are displayed on the display screen of the display unit 16. The display unit 16 then displays an image related to at least one of the design images. The display unit 16 can also display analysis results and statistical values for each design element of the design image.

[0032] First Embodiment Next, the effects of the image analysis system 1 of the first embodiment will be described. The image analysis system 1 of the first embodiment will be described by taking as an example a case where the package design of a new product that the user wishes to evaluate is analyzed. A user operates the operation input unit 14 shown in FIG. 1 to input design image data to the image analysis system 1 from the data input unit 15, the communication interface unit 13, or the storage unit 12. As shown in Fig. 2, a plurality of comparison images #1 to #3 are input as design image data together with the design image to be analyzed as an evaluation target. The comparison images are not limited to the comparison images #1 to #3 shown in Fig. 2, and a larger number of comparison images may be input.

[0033] The comparative image may be design image data of the company's own merchandise that falls into the category of the new product (merchandise) to be evaluated (e.g., chocolate, cosmetics, soft drinks, etc.). Furthermore, the comparative image may be design image data of a competitor's merchandise that competes in the same product category. Furthermore, design image data of a competitor's merchandise may be input as the comparative image together with or instead of the design image of the company's own merchandise.

[0034] The design images input to the image analysis system 1 are stored in the image database 12c of the storage unit 12. In the image analysis system 1 of this embodiment, when inputting images, the user does not need to quantitatively identify image features that contribute to the difference, including the sensory evaluation, or to previously obtain the perceptual evaluation value that serves as the explained variable.

[0035] The control unit 11 (see FIG. 1) performs arithmetic processing using a CPU. In the arithmetic processing, a program corresponding to the processing content is read from a read-only memory (ROM) 12a provided in the storage unit 12 and loaded into a random access memory (RAM) 12b. Then, the control unit 11 operates the deployed program in cooperation with various data of package images stored in the image database 12c of the storage unit 12 to generate output data to be displayed on the display unit 16. Each of the generated output data is labeled with the package image that will be the basis for analysis and stored in the storage unit 12 or output directly to the display unit 16.

[0036] When the output data is output to the display unit 16, the evaluation target image as the analysis target image and comparison images #1 to #3 are displayed in a horizontal array in the upper column of the display screen 20, as shown in Fig. 2. Furthermore, a plurality of evaluation items (for example, the attention level of a specified portion) are displayed in a vertical array in the left column of the display screen 20.

[0037] An importance column 22 is provided adjacent to the right of each evaluation item on the display screen 20. The importance column 22 displays the importance of each design element corresponding to each evaluation item. The importance varies depending on the medium to which the image to be evaluated is applied as a design. Furthermore, the display unit 16 displays numerically the evaluation value, which is the absolute value of the analysis result for each evaluation item, and the deviation value, which indicates the relative position, in the evaluation value column 23 and the deviation value column 24. Here, the deviation value is calculated based on statistical theory using the analysis results including the analysis results of the comparison image.

[0038] At this time, the control unit 11 may be configured to display a list of only the importance, evaluation value, and deviation value for each design element of the design image to be evaluated on the display screen of the display unit 16 shown in FIG. 2 as necessary.

[0039] As shown in Figure 2, the display unit 16 of the first embodiment displays, below the corresponding design image, a list of each design image, along with an evaluation value column 23 that displays the absolute evaluation value, which is the analysis result of these design images, and a deviation value column 24 that displays the deviation value calculated from the absolute evaluation value of these comparison images.

[0040] The closer the deviation value is to 50, the more similar it is to the average of the comparison images. If you want to maintain your brand image, it is recommended that you input multiple "design images of the same brand that have already been released" as comparison images, and design the image to be evaluated so that its deviation value is close to 50. If you want to differentiate your design from that of a competing brand, you can input multiple "design images of a competing brand that have already been released" and design the image to be evaluated so that its deviation value is around 40 or 60. In the first embodiment, the display screen 20 displays a list of the evaluation value column 23 and the deviation value obtained for each evaluation item.

[0041] In the image analysis system 1 of the first embodiment, a footnote column 25 and an explanation column 26 are further provided on the display screen 20 shown in FIG. The control unit 11 can display commentary information in the comment box 26 to help the user interpret the list of design images displayed on the display unit 16.

[0042] Furthermore, the user can find unique design features by looking at and interpreting the commentary information in the commentary column 26 related to the improvement policy for the image to be evaluated. Here, the commentary information includes the meaning of the displayed numerical value or information that may be generated for the viewer (consumer) of the image when the image has the characteristic indicated by the numerical value. For example, the commentary information includes information about the psychological effect on the viewer, the sensitivities evoked, or the marketing effect. In addition, the display unit 16 can also display information related to the improvement policy for the image to be evaluated as commentary information.

[0043] An example of the explanation shown in Figure 2 is, "The closer the deviation value is to 50, the more similar it is to the comparison image. If you want to maintain your brand image, it is recommended that you enter an image of a design that has already been released under the same brand as the comparison image, and make sure the deviation value is close to 50. If you want to differentiate your design from competing designs, it is recommended that you set the deviation value at around 40 or 60."

[0044] Second Embodiment Furthermore, the display unit 16 of the image analysis system 1 of the second embodiment can display, one above the other, an evaluation target image area 31 in which the user's design image is output and displayed on the display screen 30, as shown in Figure 3, and a reference image area 32 in which one or more reference images collected by the user are displayed and displayed.

[0045] The image analysis system 1 of the second embodiment acquires analysis results for each design element of a plurality of design images. The image analysis system 1 then performs principal component analysis on the analysis results for each design element to extract a first principal component and a second principal component as objective variables. The control unit 11 also performs a partial least squares regression (PLS) analysis using the quantified values for each design element as explanatory variables. Here, PLS regression analysis is an analytical method used for data containing multiple correlated explanatory variables.

[0046] The evaluation value A is calculated for each design image from the analysis results of the principal components. The following formula (1) is used to calculate the evaluation value A.

number

[0047] Furthermore, the control unit 11 of the image analysis system 1 calculates the deviation value of the evaluation target image with respect to the evaluation value A. As a result, the control unit 11 can obtain output data that can be displayed as a graph area 33 on the display screen 30 from the deviation value of the evaluation target image with respect to the calculated evaluation value A. The deviation value of the evaluation target image with respect to the evaluation value A may be temporarily stored in the storage unit 12.

[0048] The graph area 33 displays data read from the storage unit 12 by the control unit 11, or data calculated directly by the control unit 11. As a result, the graph area 33 is displayed on the display screen 30 adjacent to the evaluation target image area 31 and the reference image area 32.

[0049] The graph area 33 is preferably displayed so that the user can easily understand whether the difference between the respective design images is within a predetermined value range. The graph area 33 of the display screen 30 of the second embodiment shown in Figure 3 visually represents, using a bar graph, the analysis result of whether the difference between the evaluation target image area 31 and the reference image area 32 displayed on the display unit 16 is within a predetermined value range.

[0050] For example, the upper section of the graph area 33 displays the overall comparison analysis result 35 of the reference image and the evaluation target, which is one of the analysis results. The overall comparison analysis result 35 mainly displays a pair of hit zones 41, 42 that indicate deviation value ranges that are a predetermined distance above and below the deviation value of the reference image.

[0051] Here, the hit zone is the range that is likely to be novel compared to the reference image group, but not too unusual, and therefore most likely to attract the viewer's interest. In this case, it is expected that the effect of attracting the public's interest will be increased while ensuring that the public will recognize that the image belongs to the same category as the reference image group. The hit zone is set based on the empirical rule that when the evaluation values of the design elements of the design image to be evaluated are different by a predetermined value, the image to be evaluated is more likely to be hit. That is, the hit zone indicates a range in which the evaluation value of a certain design element is slightly different from the average for that design element in a plurality of reference images.

[0052] A black diamond mark 54 is plotted at the position of the deviation value 50 of the reference image. Furthermore, black dots 45 indicating the analysis results of the design image to be evaluated are plotted in the comprehensive comparison analysis result 35. This allows the user to visually grasp the overall position of the evaluation target relative to the distribution of multiple reference images by looking at the comprehensive comparison analysis result 35. Therefore, the user can easily recognize where the evaluation target image is positioned relative to the group of reference images displayed adjacently below.

[0053] Furthermore, for example, the center column of the graph area 33 displays quantitative analysis results 37 for each design element from the analysis results. The design elements here are the same as the evaluation items in FIG. 2, namely, blank space ratio, character area ratio, GLMC (contrast), diagonal component, average lightness, average saturation, color difference ΔE00, and base color area ratio. In the second embodiment, the quantitative analysis results for blank space ratio will be described below as a representative example. The same applies to other design elements such as character area ratio to base color area ratio, and so a description of these will be omitted.

[0054] In the quantitative analysis results 37, the values of the evaluation target image are displayed in a bold bar graph for each design element. The absolute evaluation values of the evaluation targets are values that can be calculated mechanically. Unlike the sensory evaluation given to an observer who looks at the design image, the absolute evaluation values are analytical data made up of objective numerical values that can be calculated mechanically. In the quantitative analysis results 37, the size of the unit scale on the horizontal axis showing the numerical values of the thick bar graph (for example, 56 to 76% for the margin ratio) is displayed differently for each design element.

[0055] In the quantitative analysis result 37 of the second embodiment, below the thick bar graph indicating the blank space rate, there is shown a black diamond mark 54 indicating the deviation value 50 of the reference image with respect to the blank space rate, similar zones 53 indicated by dashed lines on both sides of the black diamond mark 54, and hit zones 51 and 52 located on both sides of the similar zones 53. Therefore, according to the quantitative analysis result 37, it is possible to visually compare the absolute evaluation value of the blank space rate with the hit zones 51 and 52 displayed adjacently below, the similar zone 53, and the black diamond mark 54 indicating the deviation value 50 of the reference image. This allows the user to easily understand the overall position of the absolute evaluation value of each design element of the image to be evaluated relative to a group of multiple reference images by comparing the thick bar graph of the quantitative analysis result 37 with the black diamond mark 54 indicating the deviation value 50 of the reference image and the hit zones 51 and 52.

[0056] Therefore, by looking at the graph area 33, the user can visually and intuitively recognize the position of the image being evaluated compared to the numerical display shown in Figure 2, and can use this as information to decide which design elements should be improved. Furthermore, as shown in FIG. 3, in one example of the analysis results displayed on the display unit 16, an explanation of the overall evaluation or the evaluation of each design element is displayed in the explanation column of the graph area 33.

[0057] The user interprets the graph image by referring to the explanation displayed in the commentary section of Figure 3, which states, "If the value of the evaluation target falls within the hit zone, it is likely to be innovative compared to the reference images, but not too unusual." This allows the user to immediately recognize whether the image being analyzed is within the hit zone or not, and can reflect this in improving the design image. The user then interprets the graph display by noting, "This is expected to increase interest while ensuring recognition that the image belongs to the same category as the reference images." The user also interprets the graph display by noting, "For new product packaging, it is recommended that you modify each design element so that it falls within the hit zone." For example, a user may want to maintain the core brand image while embracing the unique design features of a particular product. In such a case, the user can minimize changes to the design image so that it falls within the area with a deviation value close to 50, which is midway between the high and low hit zones.

[0058] Then, the system may collect evaluation target images (design images that the user intends to commercialize) and comparison images (design images of competing products) limited to a specific product category, analyze the design elements of each, and display explanatory information on the display unit 16. This makes it possible to more accurately and easily determine whether the design of the image to be evaluated is a common design in the general public. Therefore, the image analysis system 1 of the second embodiment can be used to estimate the originality of a design and its ability to induce interest.

[0059] 3, the graph area 33 displays the overall comparative analysis results 35 of the reference image and the evaluation target and their legends 36. Furthermore, the quantitative analysis results 37 for each design element and their legends 38 are displayed.

[0060] Among these, legend 36 explains that the deviation value of the image to be evaluated is displayed as a black circle mark, the average deviation value of the reference image (deviation value of the reference image: 50) is displayed as a black diamond mark, the similar zone is displayed as a dashed line, and the hit zone is displayed as a solid line. In addition, legend 38 explains that the value of the image to be evaluated is shown by a hatched rectangle, the deviation value of 50 of the reference image is shown by a black diamond mark, the similarity zone is shown by a dashed line, and the hit zone is shown by a solid line.

[0061] On the display screen 30, an explanation column 39 is displayed below the graph area 33. The explanation column 39 displays an explanation of the improvement: "If the value of the evaluation target falls within the hit zone, it is likely to be novel compared to the reference image group, but not too unusual." The explanation column 39 also displays the effect of the improvement: "In this case, it is expected that the effect of attracting interest will increase while ensuring the effect of being recognized as belonging to the same category as the reference image group." This makes it easy for the user to interpret the content displayed in the graph area 33. For example, the user can learn that, depending on the design elements of the evaluation target image, it is preferable to change it so that it falls within either the hit zone 51 or 52. This allows the user to make desired changes for each design element, such as whether to place the image to be evaluated within the similarity zone to maintain the brand image, or to place it within the hit zones 51 and 52 to emphasize appropriate originality.

[0062] FIG. 4 is a flowchart illustrating the image analysis process executed by the image analysis system 1, which is a computer. The control unit 11 (see FIG. 1) of the image analysis system 1 starts image analysis processing in response to an operation instruction from the operation input unit . In step S1, the user inputs one file as an image to be evaluated. In step S2, the user inputs multiple files as reference images. In this case, the user does not need to quantitatively identify and input image features that contribute to the difference. The input file is displayed as an image to be evaluated or a reference image on the display screen of the display unit 16 shown in FIG. 3.

[0063] In step S3, the control unit 11 uses artificial intelligence to generate images similar to the reference image, thereby increasing the amount of reference image data used for analysis. Artificial intelligence can easily generate multiple similar design images. Therefore, by easily preparing and inputting multiple design images, the accuracy of the analysis can be improved. In step S4, the control unit 11 classifies the images using machine learning. Using a deep learning model that has been created by previously training a large number of images, the control unit 11 extracts image features, i.e., variables, to be used for image identification and classification for each of all images, including the image to be evaluated and the reference images. As a result, each image is vectorized and represented by multidimensional variables.

[0064] Then, in step S5, the control unit 11 performs principal component analysis on the extracted image feature amounts, and converts the multidimensional variables that represent all images, including the evaluation target image and the reference image group, into two dimensions. 5, a two-dimensional principal component analysis graph 60 plots the classification results of each image with the first principal component on the horizontal axis and the second principal component on the vertical axis. The control unit 11 can grasp the degree of variance of the first and second components from a parameter space similar to this principal component analysis graph 60.

[0065] Furthermore, in step S6, the control unit 11 sets the "first and second principal components obtained by principal component analysis" labeled for each image as the objective variables. The control unit 11 also performs PLS regression analysis using the quantified values for each design element as explanatory variables. Here, PLS regression analysis is an analytical method used for data containing multiple correlated explanatory variables.

[0066] Furthermore, in step S7, the control unit 11 outputs explanatory variables (design elements) whose importance in the PLS regression analysis is equal to or greater than a predetermined value to the display unit 16 in FIG. 1 and displays them on the display screen. 3, the design elements displayed are blank space ratio, character area ratio, GLCM (an index that expresses texture by extracting contrast components according to a predetermined rule), diagonal component, average lightness, average saturation, color difference ΔE00, and base color area ratio. Here, the control unit 11 displays the design elements in order of contribution rate.

[0067] "White space ratio" indicates the percentage of the design elements that are occupied by white space. "GLCM (contrast)" indicates a value that quantifies the appearance pattern of pixel values. It mainly increases when the granular gloss (glossiness) is high.

[0068] "Diagonal component" indicates the degree to which the objects depicted in each image are arranged diagonally. The larger the "diagonal component," the more active the image will appear. "Average brightness" indicates the overall brightness of the colors in each image. "Average saturation" indicates the overall vividness of the colors in each image.

[0069] "Color difference ΔE00" indicates the color difference between the color with the largest area ratio and the color with the second largest area ratio in each image after color reduction processing. The "base color area ratio" indicates the area ratio of the color that occupies the largest area after the color subtraction process for each image. It tends to be large when the margin ratio is large, but when fine lines are drawn in similar colors, the margin ratio is small and the base color area ratio is large.

[0070] The importance (high, medium, low) corresponding to each evaluation item, which differs depending on the design medium (in this case, packaging), is then displayed side by side.

[0071] The importance of each evaluation item is not limited to being displayed as (high, medium, low), etc. In other words, it is sufficient if the importance of the design elements is displayed qualitatively or quantitatively depending on the medium to which the user applies the design image being evaluated. The medium is not limited to packaging, but may also be, for example, a website, product shelf layout, POP, television commercials, etc. The user can consider the weighting by looking at the display of the importance of the design elements.

[0072] In step S8, the control unit 11 calculates an evaluation value A from the first and second principal components for each evaluation target image and each reference image based on the principal component analysis results in step S5. In step S9, the control unit 11 displays on the display unit 16 the deviation value of the image to be evaluated for the evaluation value A, the deviation value 50 of the reference image for the evaluation value A (average of the reference image group), and the range in which the deviation value of the evaluation value A takes on a predetermined value.

[0073] Then, the display unit 16 displays the analysis results and statistical values in the graph area 33 shown in FIG. 3, and the image analysis process of the image analysis system 1 is completed.

[0074] As shown in FIG. 3, the deviation value of the image to be evaluated for the evaluation value A calculated by formula (1), the deviation value indicating the average of the reference images for the evaluation value A, and the range in which the deviation value of the evaluation value A takes on a predetermined value are displayed in the comprehensive comparison analysis result 35 in the graph area 33. The image analysis system 1 of the second embodiment converts each design element into an absolute evaluation value, performs principal component analysis on the absolute evaluation values, and displays the design elements in order of contribution rate. By displaying each design element in order of contribution rate, it is possible to show the user how explanatory each design element is in classifying images. On the display unit 16, a comprehensive comparison analysis result 35 of the reference image and the image to be evaluated is displayed in the upper column of the graph area 33 (see FIG. 3). In the comprehensive comparison analysis result 35 of the second embodiment, the range of hit zone 41 indicating the low deviation value area is within the range of deviation values from 45 to 48. Furthermore, the range of hit zone 42 indicating the high deviation value area is within the range of deviation values from 52 to 55. These hit zones 41 and 42 are displayed by solid lines.

[0075] Furthermore, a range sandwiched between these hit zones 41 and 42 and resembling the average of the reference image group is displayed by a dashed line as a similarity zone 43 . A black diamond mark 44 is plotted as the reference image average in the middle of the similarity zone 43. The deviation value of the reference image average is 50. Furthermore, a black circle mark 45 indicating the result of the analysis of the image to be evaluated is plotted within the hit zone 42. Since the black circle mark 45 of the result of the analysis of the image to be evaluated in the second embodiment is within the hit zone 42, it can be seen that the result of the comprehensive analysis is "highly likely to be innovative, but not too unusual."

[0076] In this way, the system performs principal component analysis on multiple design elements of multiple design images and displays the design elements in order of their contribution rate, allowing the user to intuitively know which design element is the determining factor in determining the differences between the multiple images. Therefore, the image analysis system 1 of the embodiment can provide practically beneficial effects such as being able to easily and effectively propose improvements to design images.

[0077] As described above, the image analysis system 1 of the second embodiment includes a control unit 11 that acquires analysis results obtained by analyzing each design element of each of a plurality of design images and calculates statistical values using the analysis results. The image analysis system 1 also includes a display unit 16 that displays images related to at least one of the design images, as well as the analysis results and statistical values for each design element of the design image. This provides an image analysis system that can easily and effectively propose improvements to design images.

[0078] Specifically, multiple design images can be collected and analyzed simultaneously for each design element. Users do not need to quantitatively identify and input image features that contribute to differences, which reduces the effort required for input. The analysis results and statistical values using the analysis results are displayed together with an image related to the design image on the display unit 16. This allows the analysis results and statistical values to be compared with the analyzed design image.

[0079] Furthermore, the feature quantities of a design image can be compared with other design images for each design element, making it possible to easily find unique design features. Therefore, the present invention displays the meaning of the numerical values of the analysis results output for each design element for the design image entered by the user, as well as design improvement policies in an easy-to-understand manner, allowing the user to easily and effectively propose improvements to the design image and even determine the improvement policies.

[0080] The control unit 11 also calculates statistical values using objective analysis results obtained by mechanically analyzing multiple design images. Since the statistical values are obtained by statistically processing mechanically calculated absolute evaluation values without taking into account sensory evaluation, users can make objective judgments. For example, it becomes easier to determine whether the identified design features form the framework for the product's brand image or are elements that differentiate it from competing designs. This makes it easier for users to formulate design strategies. Furthermore, for example, a user can easily perform analysis by collecting and inputting multiple design images for a specific product. This allows the user to reduce the amount of work required to obtain affective evaluation value data. Furthermore, the user can easily find design features unique to a specific product, allowing them to effectively improve or propose improvements to the design images.

[0081] Furthermore, the statistical values of the image analysis system 1 include deviation values calculated using the analysis results obtained by analyzing the plurality of design images for each design element based on statistical theory. Therefore, the display unit 16 displays the feature amount of the design image for each design element using a deviation value. This allows the user to quantitatively evaluate the difference in features between the image to be evaluated and the reference image. Therefore, if the user wants to increase or decrease the difference between the image to be evaluated and the reference image, the user can easily determine which design element feature should be changed to be effective. This makes it possible to more easily and effectively propose improvements to the design image.

[0082] Furthermore, the display unit 16 displays a list of the plurality of design images for which the analysis results have been acquired, as well as the analysis results and statistical values analyzed from the design elements of each design image. This allows the display unit 16 to display a list of analysis results and statistical values for each design image. This allows a quantitative comparison to be made by visually comparing the evaluation target image with a comparison image, which is another design image. This allows the user to easily determine which design element features should be changed to be effective. This makes it even easier to make effective improvement suggestions for design images.

[0083] Furthermore, the control unit 11 causes the explanation columns 26 and 39 of the display unit 16 to display commentary information for the user to interpret the list displayed on the display unit 16 . Therefore, when comparing the analysis results and statistical values of each design image displayed in a list on the display unit 16, the user can, for example, interpret and find the unique design features while looking at explanatory information related to the improvement policy for the image being evaluated. Here, the commentary information includes information about the meaning of the displayed numerical value, or the psychological effect, sensibility, or marketing effect that may occur in the observer (consumer) of the image when the image has the characteristics indicated by the numerical value.In addition, the commentary information can also display information related to the improvement policy for the image to be evaluated.

[0084] For example, one example of the explanation shown in Figure 2 is, "The closer the deviation value is to 50, the more similar it is to the comparison image. If you want to maintain your brand image, it is recommended that you enter an image of a design that has already been released under the same brand as the comparison image, so that the deviation value is as close to 50 as possible."

[0085] Such marketing-related information is useful when you want to maintain a brand image. For example, you might want to maintain a brand image through design images, and make a poster for cosmetics company S more easily convey the unique characteristics of cosmetics company S. That is, the user inputs a publicly known design image of the same brand as a comparison image. Then, the user changes the deviation value for each design element so that it approaches 50. By weighting the importance of the design elements, the user can easily bring the deviation value close to 50.

[0086] Furthermore, for example, in the case of a package for soft drink company C, it is easy to make improvement suggestions according to the purpose, such as maintaining the design image of the main soft drink C1 while differentiating it from a new soft drink C2. If you want to differentiate your design from competing designs, we recommend setting the deviation value around 40 or 60. (These recommended deviation values are current recommendations and may change as the system improves.)

[0087] The display unit 16 can then display the importance of the design elements qualitatively or quantitatively depending on the medium to which the design image that the user has selected as the evaluation target is to be applied. Here, media includes packaging, websites, shelf layouts, point-of-purchase advertising (promotional materials used to display products on shelves), television commercials, etc. For example, in the case of packaging, the margin ratio and character area ratio are considered to be of high importance, while the diagonal (135 degree) component is considered to be of low importance. In the case of packaging, the design technique of placing pictures diagonally is considered to be of low importance. In the case of a website, the margin ratio and character area ratio are of "medium" importance, and the diagonal (135 degrees) component is of "high" importance. In this way, the image analysis system 1 displays the importance of each design element in a stepwise or non-stepwise manner, from high to low, etc. This makes it possible to make optimal design image improvement suggestions depending on the medium.

[0088] The display unit 16 can also display analytically processed data or other data extracted from the multiple images. Furthermore, the display unit 16 can display variables with high explanatory power from variables included in the image-related data or variables included in the non-image-related data for the data analysis results using at least one of the integrated data. That is, the variables are not limited to image feature amounts, and may be any variables related to design elements including character information. For example, the diagonal component (135-degree component) as a design element shown in Fig. 2 has a "low" importance level, and therefore may be considered a variable (evaluation item) with low explanatory power and may not be displayed on the display unit 16. In addition, the variable may be configured to display other variables related to design elements, including image features or text information, with high explanatory power.

[0089] As a result, the analysis results and the statistical values displayed on the display unit 16 are further organized, and the design elements of the design image are carefully selected to have variables with high explanatory power (i.e., importance, contribution rate, etc.). This makes it easier to compare with other design images, find unique design features, and make effective suggestions for improving the design image.

[0090] The display unit 16 then displays, in the graph area 33, the area where the difference between the image to be evaluated and the reference image is within a predetermined range. This allows the user to visually and intuitively recognize whether the difference between the evaluation target image area 31 and the reference image area 32 displayed on the display unit 16 is within a specified value range, and use this as information for making a decision.

[0091] For example, as shown in FIG. 3, in the graph area 33, quantitative analysis results 37 for each design element are displayed in the middle column below the overall comparative analysis results 35. In the quantitative analysis result 37 of the second embodiment, for example, in the blank space ratio category of the design elements, hit zone 51, which indicates a low deviation value area, is in the blank space ratio range of 63 to 65.5. Also, hit zone 52, which indicates a high deviation value area, is in the blank space ratio range of 69 to 71.8, as indicated by a solid line.

[0092] Furthermore, a range that is sandwiched between these hit zones 51 and 52 and that resembles the comparison image is displayed by a dashed line as a similarity zone 53. A black diamond mark 54 is plotted in the middle of the similarity zone 53 as a deviation value indicating the average of the reference images. In the comprehensive comparison analysis result 35 of the second embodiment, the black diamond mark 54 of the deviation value indicating the average of the reference images indicates that the blank space ratio is 67.5. In other words, it can be seen that in order to generate a design image with a style similar to that of the reference image group, it is sufficient to set the blank space ratio to around 67.5. Furthermore, the absolute evaluation value 55 is displayed as a bold bar graph. For the blank space ratio, the absolute evaluation value 55 is "73." It is easy to see that the absolute evaluation value 55 for the blank space ratio is outside the hit zones 51 and 52.

[0093] In addition to the overall comparative analysis results 35 shown in the graph area 33, the user can visually and intuitively recognize the position of the absolute evaluation value 55 of each evaluation item, which can be used as information for making decisions about improvements. Additionally, an explanation column 39 is displayed in the lower section of the same graph area 33. This allows the user to easily implement the best improvement method, such as "first, bring the margin ratio into the range of the similarity zone 53," based on the advice in the explanation column 39 that "...for new product packaging, etc., it is recommended that you modify each design element so that it falls within the hit zone." Other evaluation items (such as character area ratio) are similar to the blank space ratio, so their explanation will be omitted.

[0094] In this way, principal component analysis is performed on multiple design images and the design elements are displayed in order of contribution rate, so the user can intuitively know the design element that is the deciding factor in determining the differences between multiple images.

[0095] Therefore, the image analysis system 1 of the embodiment can provide practically beneficial effects such as being able to easily and effectively propose improvements to design images. In one example of the analysis results, in addition to an explanation column 39 and comprehensive comparative analysis results 35, quantitative analysis results 37 are displayed. These quantitative analysis results 37 show the absolute evaluation value of the image being evaluated for each evaluation item in a bar graph, and for each item, a black diamond mark indicating a deviation value of 50 for the reference image group and a pair of high and low hit zones located on either side of it are displayed.

[0096] The user interprets these graph images by referring to the explanation in the commentary section 39 of Figure 3, which states, "If the evaluation target value falls within the hit zone, it is likely to be innovative and not too unusual compared to the reference images." This allows the user to immediately recognize whether the image being analyzed is within the hit zone. Furthermore, for items outside the hit zone, guidance is easily provided on how to improve the design image.

[0097] The user also reads, "In this case, it is expected that the effect of attracting attention will be increased while ensuring recognition that the image belongs to the same category as the reference images." The user also reads, "For new product packaging, it is recommended that each design element be modified so that it falls within the hit zone." The user then interprets the graph display. For example, a company may want to maintain the framework of a brand image while embracing the unique design features of a particular product. In such a case, the user can minimize changes to the design image so that it falls within the area with a deviation value close to 50, which is midway between the high and low hit zones.

[0098] Therefore, the control unit 11 collects evaluation target images (design images that the user intends to use on product packaging) and reference images (design images of product packaging in the same category) limited to a specific product category, analyzes the design elements of each, and displays explanatory information on the display unit 16. This makes it possible to determine whether the design of the image to be evaluated is a common design among the general public. Therefore, the image analysis system 1 can be used to estimate the originality of a design and its ability to induce interest. In addition, by collecting past company design images of the same product as reference images, and analyzing and displaying them as in Figure 2 and Figure 3, it can be determined that the closer the image being evaluated is to the reference image, the better the brand image has been maintained. This can be used as an objective indicator when considering the design of products for which maintaining brand image is important.

[0099] In the second embodiment, when analyzing the design elements of a design image, the control unit 11 generates similar design images using artificial intelligence (AI). Then, the control unit 11 adds the generated similar design images to analyze the design elements.

[0100] Artificial intelligence can easily generate a large number of similar design images, so the image analysis system 1 can prepare and analyze a large number of comparative design images, further improving the accuracy of the analysis.

[0101] In this way, the image analysis system 1 described in the second embodiment of the present invention can compare the features between images by design element, so that if a user collects design images that are based on a specific product and analyzes them simultaneously, they can easily find the design features unique to that product. This makes it easier to determine whether the features identified form the backbone of the product's brand image or whether they are elements that differentiate it from competing designs, making it easier to formulate a design strategy.

[0102] Furthermore, since the difference between the features of the image to be evaluated and the reference image can be quantitatively evaluated, the user can easily determine which features to change if they want to increase or decrease the difference between the image to be evaluated and the reference image.

[0103] Furthermore, the image analysis system can achieve practically beneficial effects, such as improving the analytical accuracy for calculating deviation values, differences, and the like, which are shown as relative evaluation values.

[0104] Although the embodiments and their modifications of the present invention have been described above, these embodiments are merely illustrative and do not limit the technical scope of the present invention. The present invention can take on various other embodiments, and various modifications such as omissions and substitutions can be made without departing from the spirit of the present invention. These embodiments and their modifications are included within the scope and spirit of the invention described in this specification, etc., and are included in the invention described in the claims and their equivalents.

[0105] For example, in the first embodiment, the display screen 20 shown in FIG. 2 displays the numerical values such as the deviation value on the display unit 16, but the present invention is not limited to this. For example, as shown in Fig. 3 of the second embodiment, the calculated deviation values and the like may simply be displayed as a graph image. In other words, as long as the analysis results and statistical values for each design element of the design image are displayed on the display unit 16, the number, shape, layout, and type of graph to be displayed are not particularly limited.

[0106] In addition, for example, in the above embodiment, the statistical value is a deviation value calculated based on statistical theory, but this is not limited to this. For example, the statistical value may be a combination of a standard deviation, an average value, a median value, or the like with a deviation value. In other words, the statistical values displayed on the display unit may be displayed together with the image related to the design image and the analysis results. Therefore, the number, type, and display format of the displayed statistical values are not particularly limited.

[0107] Furthermore, in the image analysis system 1 of the above embodiment, a PLS regression analysis is performed after obtaining quantifiable values using principal component analysis. However, the present invention is not limited to this, and any analysis method may be used as long as it can obtain analysis results for each design element. [Explanation of symbols]

[0108] 1. Image analysis system 11 Control section 12 Storage section 13 Communication interface section 14 Operation input section 15 Data Entry Section 16 Display section 12a ROM 12b RAM 12c Image Database 12d Image Analysis Program 20 display screen 22 Importance column 23 Evaluation value column 24 Standard deviation column 25 Footnote column 26,39 Commentary

Claims

1. a control unit that acquires analysis results obtained by analyzing each of the plurality of design images for each design element and calculates statistical values of the analysis results; and a display unit that displays at least one of the design images, the analysis results of the design image, and the statistical values of the design image.

2. The image analysis system according to claim 1 , wherein the statistical values include any one of deviation, standard deviation, mean value, and median value of analysis results obtained by analyzing a plurality of the design images for each design element.

3. 3. The image analysis system according to claim 1, wherein the display unit displays a list of a plurality of design images from which the analysis results have been obtained, and the analysis results and statistical values thereof analyzed from the design elements of each of the design images.

4. The image analysis system according to claim 3 , wherein the control unit causes the display unit to display commentary information for a user to interpret the list displayed on the display unit.

5. The image analysis system of claim 4, wherein the explanatory information is information regarding the meaning of the displayed numerical value, or any of the psychological effects, emotions, or marketing effects that may occur in an observer of the image when the image has the characteristics indicated by the numerical value.

6. The image analysis system according to claim 1 , wherein the display unit displays the importance of the design elements qualitatively or quantitatively depending on a medium to which the design image evaluated by the user is applied.

7. The image analysis system of claim 1, wherein the display unit displays variables with high explanatory power from variables contained in image-related data or variables contained in related data other than images for data analysis results using at least one of analytical processing data, other data extracted from multiple design images, or data integrating these.

8. The image analysis system according to claim 4 , wherein the display unit displays, as a graph image, an area where the difference between the evaluation target image and the comparison image is within a predetermined range.

9. The image analysis system of claim 1 , wherein the control unit, when analyzing design elements of a design image, generates the similar design image using artificial intelligence and adds the generated similar design image to analyze the design elements.

10. On the computer, A step of acquiring an analysis result obtained by analyzing each of the plurality of design images for each design element; calculating statistics of the analysis results; displaying at least one of the design images together with the analysis results and the statistics for that design image; Image analysis program to run.

11. acquiring an analysis result obtained by analyzing each of the plurality of design images for each design element; calculating statistics of the analysis results; displaying at least one of the design images together with the analysis results and the statistics for that design image; An image analysis method comprising:

Images