Foot pronation estimation device, foot pronation estimation method, and foot pronation estimation program

Abstract

To provide a technique for easily determining a pronation type.SOLUTION: A foot pronation estimation device 50 is configured so that: a pronation estimation part 72 estimates a degree of foot pronation on the basis of comparison between a reference value and a feature amount in a prescribed portion; a feature amount extraction part 60 extracts feature amounts of multiple portions including at least, a length of a foot width, and a recess amount in a shape of a shank part, from top view images. The pronation estimation part 72 estimates the degree of the foot pronation on the basis of comparison between the reference values, and the feature amounts of the multiple portions including at least the length of the foot width and the recess amount of the shape of the shank part. The feature amounts of the multiple portions further include a protrusion amount in an outline shape in the vicinity of a small toe thenar and a foot length. The pronation estimation part 72 estimates which of a first type where the pronation degree is higher than a predetermined reference and a second type where the pronation degree is equal to or less than the predetermined reference corresponds to as a type classified according to the degree of the foot pronation.SELECTED DRAWING: Figure 2

Description

[Technical field]

[0001] The present disclosure relates to a foot pronation estimation device, and more particularly to a technique for analyzing a foot condition from a foot shape image. [Background technology]

[0002] Some running shoe products are categorized according to the type of pronation. The categories are often divided into those for runners who overpronate and those for others (neutral pronation, underpronation). Runners can maintain appropriate pronation and reduce the risk of injury by using running shoes in a category that is appropriate for their pronation type.

[0003] There is known a technique for measuring the footprint or pressure distribution of the foot to estimate the arch height ratio or foot flexibility of a customer and selecting a type of shoe that fits the customer (see, for example, Patent Document 1). [Prior art documents] [Patent documents]

[0004] [Patent Document 1] WO 05 / 006905 Summary of the Invention [Problem to be solved by the invention]

[0005] In order to select running shoes according to the pronation type, it is necessary to be able to recognize which pronation type one is. However, in order to determine the pronation type, it is usually necessary to photograph the running on a treadmill from behind and analyze the foot movement, and it is necessary to use a facility that has the equipment for such photography and analysis. Alternatively, the pronation type determination technology in Patent Document 1 can estimate the pronation type from the footprint and the pressure distribution of the foot, but since it requires the use of measuring equipment such as an optical sensor or a pressure sensor, it is not necessarily easy for general runners to determine the pronation type.

[0006] On the other hand, since shoe manufacturers categorize running shoes according to pronation type, it is desirable to avoid type mismatches, such as runners who do not overpronate using shoes for overpronation. In particular, in recent years, when online shoe sales have become common, it has been desirable to establish a method to easily determine pronation type even in environments without dedicated equipment.

[0007] The present disclosure has been made in consideration of these circumstances, and has a purpose to provide a technique for easily determining the pronation type. [Means for solving the problem]

[0008] In order to solve the above problems, a foot pronation estimation device according to one aspect of the present disclosure includes an image acquisition unit that acquires at least a top-view image of the foot as an image showing the contour shape of the subject's foot when viewed from above, a feature extraction unit that extracts features of specific parts that indicate the top-view contour shape of the foot from the top-view image, and a pronation estimation unit that estimates the degree of foot pronation based on a comparison between the features of the specific part and a reference value.

[0009] Another aspect of the present disclosure is a method for estimating foot pronation, comprising the steps of: acquiring at least a top-view image of the foot of a subject as an image showing a contour shape of the foot in a top view; extracting, from the top-view image, feature amounts of a predetermined portion showing the contour shape of the foot in a top view by image processing; and estimating a degree of foot pronation based on a comparison process between the feature amounts of the predetermined portion and a reference value. Effect of the Invention

[0010] According to the present disclosure, a technique for easily determining the pronation type can be provided. [Brief description of the drawings]

[0011] [Figure 1] FIG. 1 is a diagram showing a basic configuration of a foot pronation estimation system. [Diagram 2] FIG. 2 is a functional block diagram showing each component of the foot pronation estimation device. [Diagram 3] FIG. 1 is a diagram showing a method for measuring foot length and foot width. [Figure 4] FIG. 4 is a diagram showing a method for measuring the amount of inner recession and the amount of outer protrusion. [Diagram 5] FIG. 13 is a diagram showing the relationship between pronation type and shoe type. [Figure 6] FIG. 13 is a diagram showing the results of matching the classification results of subjects by decision tree analysis with the classification results of pronation types based on three-dimensional valgus angles. [Figure 7] FIG. 13 is a tree diagram that illustrates the results of a decision tree analysis of foot pronation estimation based on three elements: foot length, foot width, and medial concavity. [Figure 8] 13 is a flowchart showing the process of foot pronation estimation processing using a foot pronation estimation algorithm based on three elements: foot length, foot width, and medial concavity amount. [Figure 9] FIG. 13 is a tree diagram that illustrates the results of a decision tree analysis of foot pronation estimation based on three elements: foot width, medial concavity, and lateral protrusion. [Figure 10] 13 is a flowchart showing the process of foot pronation estimation processing using a foot pronation estimation algorithm based on three elements: foot width, amount of medial depression, and amount of lateral protrusion. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0012] Hereinafter, the embodiments will be described with reference to the drawings. In the embodiments and modifications, the same or equivalent components are denoted by the same reference numerals, and redundant descriptions will be omitted as appropriate.

[0013] In this embodiment, a feature value indicating the top view contour shape of the foot is extracted from a top view image of the foot of a subject in a sitting position, and the degree of pronation of the foot is estimated based on a comparison with a reference value. The top view image is an image indicating the contour shape of the foot of the subject when viewed from above. Since the top view image includes the contour shape of the entire foot, such as the shape of the toes and the shape of the heel, the feature values ​​of the foot parts, such as the foot length and width, can be extracted from the top view image. As a method for estimating the pronation type based on the feature values ​​of the foot parts, a method is adopted in which the relationship between the feature values ​​of the foot parts measured for multiple subjects and the pronation type is analyzed in advance using a decision tree, and an estimation algorithm based on the analysis results is used to estimate the type.

[0014] 1 shows a basic configuration of a foot pronation estimation system 100. The foot pronation estimation system 100 includes a device for acquiring a top-view image of the foot, such as an information terminal 16 or a three-dimensional foot shape measuring device 18, and a foot pronation estimation device 50. For example, a runner who is considering purchasing running shoes categorized according to pronation type is taken as a measurement subject 10, and it is assumed that the measurement subject 10 uses the information terminal 16 or the three-dimensional foot shape measuring device 18 to determine the pronation type of his / her own foot. The measurement subject 10 acquires a top-view image of each foot and determines the pronation type for each of the left and right feet.

[0015] The information terminal 16 may be a portable information terminal such as a smartphone or a tablet terminal, or may be a personal computer. The information terminal 16 is configured by a combination of hardware such as a camera module, a distance measuring sensor, a microprocessor, a touch panel, a memory, and a communication module. The information terminal 16 may be a general-purpose information terminal such as a smartphone or a tablet terminal provided by the subject 10 or a shoe store. The information terminal 16 accesses a website provided by the foot pronation estimation device 50 via a web browser, or displays information provided by the foot pronation estimation device 50 by application software running on the information terminal 16.

[0016] When using the information terminal 16, the subject 10 takes an image of his / her foot in a sitting position using a camera built into the information terminal 16, such as his / her mobile phone terminal, to obtain an image. For example, when taking an image of the entire foot from directly above the foot, it is difficult to capture the outline shape of the entire foot in one image, because the ankle is also captured even if the heel is captured sufficiently. Therefore, it may be assumed that the image is taken so that at least the outline of the toes and heel, and the outlines of both the left and right sides from the forefoot to the midfoot are captured.

[0017] If the information terminal 16 has a three-dimensional scanner function using technology such as LiDAR (Light Detection And Ranging), the surroundings of the foot may be scanned to generate a three-dimensional model of the foot shape, and a top-view image, which is a two-dimensional image, may be obtained from the three-dimensional model. Even if the information terminal 16 does not have a three-dimensional scanner function using LiDAR or the like, a three-dimensional model of the foot shape may be generated by image synthesis processing such as photogrammetry, and a top-view image, which is a two-dimensional image, may be obtained from the three-dimensional model. The subject 10 may place his / her foot on a dedicated measurement mat 12 and scan his / her own foot shape using the camera function and foot shape acquisition application of the information terminal 16 to generate a three-dimensional model of the foot shape.

[0018] When using the three-dimensional foot shape measuring device 18, the subject 10 may generate a three-dimensional shape of the foot using the three-dimensional foot shape measuring device 18 installed at a shoe store, for example, and obtain a top view image, which is a two-dimensional image, from the three-dimensional model. The three-dimensional foot shape measuring device 18 obtains three-dimensional data of the foot shape by laser measurement. The measurement value as the three-dimensional measurement result of the foot shape scanned by the three-dimensional foot shape measuring device 18 is transmitted from the three-dimensional foot shape measuring device 18 to the foot pronation estimation device 50, and the foot pronation estimation device 50 may generate a top view image, which is a two-dimensional image, from the three-dimensional model.

[0019] In a modified example, a side view image may be acquired instead of a top view image of the feet of the subject 10. Alternatively, both a top view image and a side view image of the feet of the subject 10 may be acquired. Alternatively, a top view image or a side view image of the feet of the subject 10 in a standing position may be acquired.

[0020] The foot pronation estimation device 50 is connected to a plurality of information terminals 16 or a three-dimensional foot shape measuring device 18 via a network line such as the Internet or a LAN (Local Area Network) or a communication means such as wireless communication. The foot pronation estimation device 50 may be realized as a server that estimates the pronation type of the foot based on the measurement data transmitted from the plurality of information terminals 16 or the three-dimensional foot shape measuring device 18. The foot pronation estimation device 50 may be configured as a single server computer or a combination of a plurality of server computers. The "foot pronation estimation device" in the claims may refer to the entire foot pronation estimation system 100 or may refer to the foot pronation estimation device 50. In the present embodiment, the foot pronation estimation device 50 is realized in a form in which many of the characteristic functions included in the "foot pronation estimation device" in the claims are provided, so that the foot pronation estimation device 50 essentially corresponds to the "foot pronation estimation device" in the claims. However, in a modified example, the information terminal 16 may be configured to perform processes such as obtaining top view images, extracting foot part features, and outputting the foot pronation type estimation results, and the "foot pronation estimation device" referred to in the claims may be realized by collaboration between the foot pronation estimation device 50 and the information terminal 16.

[0021] FIG. 2 is a functional block diagram showing each configuration of the foot pronation estimation device 50. FIG. 2 illustrates functional blocks that can be realized by cooperation of various hardware configurations and software configurations regarding the foot pronation estimation device 50. Those skilled in the art will understand that these functional blocks can be realized in various forms by hardware only, software only, or a combination thereof. The foot pronation estimation device 50 is composed of a combination of hardware such as a microprocessor, a memory, a display, and a communication module. The foot pronation estimation device 50 may be a server computer built and managed by a shoe manufacturer. However, the foot pronation estimation device 50 runs a program having the following functions. The foot pronation estimation device 50 includes, as its functions, a communication unit 52, an image acquisition unit 54, a feature extraction unit 60, an evaluation unit 70, a memory unit 80, an output unit 90, and an analysis processing unit 99. The communication unit 52 of the foot pronation estimation device 50 and the information terminal 16 or the three-dimensional foot shape measurement device 18 are connected via a network.

[0022] The image acquisition unit 54 acquires the top view image transmitted from the information terminal 16 or the three-dimensional foot shape measuring device 18 via the communication unit 52. The image acquisition unit 54 acquires an image showing the top view contour shape of the foot of the subject 10 in a sitting position as the top view image. The image acquisition unit 54 acquires a top view image for each of the left and right feet. However, it is not essential to acquire images of both the left and right feet, and an image of either the left or right foot may be acquired. In a modified example, the image acquisition unit 54 may acquire a top view image and a side view image of the foot of the subject 10. In another modified example, the image acquisition unit 54 may acquire a top view image or a side view image of the foot in a standing position.

[0023] The image acquisition unit 54 includes a measurement value acquisition unit 56 and an image generation unit 58. When three-dimensional measurement values ​​of the foot shape are transmitted from the information terminal 16 or the three-dimensional foot shape measurement device 18 instead of a top-view image, the measurement value acquisition unit 56 acquires the three-dimensional measurement values ​​via the communication unit 52. In this case, the image generation unit 58 generates a three-dimensional model based on the three-dimensional measurement values, and generates a top-view image, which is a two-dimensional image, from the three-dimensional model. The technology for generating a three-dimensional model from three-dimensional measurement values ​​is well known, so a description thereof will be omitted.

[0024] The feature amount extraction unit 60 extracts feature amounts of a plurality of foot parts from the top view image. The feature amounts of the plurality of foot parts are at least one of (1) foot length, (2) foot width, (3) indentation amount in the shape of the arch (hereinafter referred to as "inner indentation amount"), and (4) protrusion amount in the contour shape near the ball of the foot (hereinafter referred to as "outer protrusion amount"). The feature amount extraction unit 60 includes a foot length calculation unit 62, a foot width calculation unit 64, a indentation amount calculation unit 66, and a protrusion amount calculation unit 68. The foot length calculation unit 62 calculates the foot length from the top view image. The foot width calculation unit 64 calculates the foot width from the top view image. The indentation amount calculation unit 66 calculates the inner indentation amount from the top view image. The protrusion amount calculation unit 68 calculates the outer protrusion amount from the top view image.

[0025] The pronation estimation unit 72 estimates the degree of pronation of the foot based on a comparison between a feature amount including at least one of the foot length, foot width, medial indentation amount, and lateral protrusion amount and a reference value. The pronation estimation unit 72 estimates which of a plurality of pronation types the foot of the subject 10 corresponds to, using a foot pronation estimation algorithm determined in advance in the design stage by a decision tree analysis described later. The pronation estimation unit 72 estimates the pronation type for each of the left and right feet of the subject 10.

[0026] The storage unit 80 includes a reference storage unit 82, a pronation type storage unit 84, and a shoe type storage unit 86. The reference storage unit 82 stores reference values ​​for each of the feature amounts of a plurality of foot parts, which are determined by a decision tree analysis described below.

[0027] The pronation type storage unit 84 stores a plurality of pronation types classified according to the degree of pronation of the foot, including at least a first type in which the degree of pronation is higher than a predetermined standard and a second type in which the degree of pronation is equal to or lower than a predetermined standard. The predetermined standard here is, for example, "neutral pronation" in which the heel moderately collapses inward when landing.

[0028] The shoe type memory unit 86 stores information on shoe product models, classified according to how the shoe types deal with the pronation of the foot, into a first shoe type for dealing with a first type in which the degree of pronation is higher than a predetermined standard, and a second shoe type for dealing with a second type in which the degree of pronation is equal to or lower than the predetermined standard.

[0029] The evaluation unit 70 includes a pronation estimation unit 72, a type determination unit 74, and a shoe type storage unit 86. The pronation estimation unit 72 estimates the pronation type of the foot of the subject 10 using a foot pronation estimation algorithm described below. The type determination unit 74 determines a shoe type to be recommended to the subject 10 from a plurality of shoe types based on the degree of pronation estimated by the pronation estimation unit 72. The shoe selection unit 76 selects a shoe product model corresponding to the shoe type determined by the type determination unit 74 from the shoe type storage unit 86.

[0030] The output unit 90 includes a result output unit 92 and a recommendation output unit 94. The result output unit 92 transmits information on the pronation type of the foot of the subject 10 determined by the pronation estimation unit 72 and information on the shoe type suited to the foot of the subject 10 determined by the type determination unit 74 to the information terminal 16 via the communication unit 52. The recommendation output unit 94 transmits information on the shoe product model selected by the shoe selection unit 76 to be recommended to the subject 10 to the information terminal 16 via the communication unit 52.

[0031] The analysis processing unit 99 determines a foot pronation estimation algorithm for estimating whether the degree of foot pronation corresponds to the first type or the second type based on a plurality of feature amounts measured from foot images of a plurality of subjects during the design stage of the foot pronation estimation device 50. The analysis processing unit 99 determines the foot pronation estimation algorithm by decision tree analysis, particularly a regression tree. The decision tree analysis is performed in advance at a stage prior to the time of estimating foot pronation for the measurement subject 10, for example, at the design stage of the foot pronation estimation system 100. The judgment procedure and judgment threshold for each feature amount determined by the decision tree analysis are adopted as a foot pronation estimation algorithm in the foot pronation estimation process in the pronation estimation unit 72. The judgment threshold for each feature amount determined by the decision tree analysis is stored in the reference storage unit 82 as a judgment reference value for each feature amount in the foot pronation estimation process. Candidates for feature amounts of each foot part used for estimating foot pronation are foot length, foot width, medial concavity amount, and lateral protrusion amount, all of which are feature amounts that can be extracted from a top view image.

[0032] The decision tree analysis in the design stage is carried out in the following procedure. First, the three-dimensional heel valgus angle is detected from a video of the subject running barefoot, captured by a motion capture system, and the actual pronation type is determined. In particular, if the heel valgus angle is greater than a predetermined threshold value X, it is determined to be overpronation or the first type. Note that a method of detecting a two-dimensional heel valgus angle from a video captured from the rear of the subject running barefoot on a treadmill may also be adopted. Also, even if the subject is not barefoot, the subject may run while wearing shoes that do not have a pronation suppression function. Such a pronation type determination is carried out for n subjects, and the subjects are classified into pronation types. Meanwhile, the feature values ​​of multiple foot parts are measured as target data for the decision tree analysis for these subjects. For example, foot length, foot width, amount of medial concavity, and amount of lateral protrusion are extracted from a top view image of the foot as the feature values ​​of the foot parts.

[0033] A decision tree analysis is performed with the classification of n subjects as to whether they are overpronating or not as the objective variable, and two or more elements of foot length, foot width, medial indentation, and lateral protrusion as explanatory variables. The decision tree analysis is performed for all combination patterns while changing the combination of explanatory variables, such as a combination of two elements, a combination of three elements, and a combination of four elements, out of the four elements of foot length, foot width, medial indentation, and lateral protrusion. In the decision tree analysis, the maximum node depth is set to 5, and the explanatory variable element and its threshold value that minimizes the squared error, which is the loss function for each branched node, are regressively determined.

[0034] The classification result of the subject by the decision tree analysis is compared with the classification result of the pronation type based on the three-dimensional valgus angle. When the classification results of all the subjects match 100% as a result of the comparison, the combination of the features in the case where the combination is an effective explanatory variable that can estimate the foot pronation with high accuracy. The branching condition and the branching order in the decision tree analysis that can estimate the foot pronation with high accuracy are adopted in the foot pronation estimation algorithm used in the pronation estimation unit 72. The threshold value for each feature determined as the branching condition is stored in the reference storage unit 82 as the reference value for each feature in the foot pronation estimation algorithm. The reference storage unit 82 may store not only the reference value for each feature, but also the foot pronation estimation algorithm itself. For example, when an algorithm for determining the foot pronation type with higher accuracy and a reference value for each feature are determined by the decision tree analysis through a post-hoc additional experiment, the information stored in the reference storage unit 82 may be rewritten. This allows the foot pronation estimation algorithm and the reference value for each feature to be updated, and the estimation accuracy to be easily improved.

[0035] 3 shows a method for measuring the foot length and foot width. The feature extraction unit 60 sets a bounding box 21 surrounding the contour of the foot 20 in the top view image 30. The bounding box 21 is a rectangular area set to be the smallest rectangle that circumscribes the entire contour of the foot 20. The points of contact between the contour of the foot 20 and the bounding box 21 are the toe side end point 24, the heel side end point 25, the inner end point 28, and the outer end point 29.

[0036] The sole of the foot other than the toes is mainly divided into a forefoot portion 40, a midfoot portion 41, and a rearfoot portion 42. The area of ​​the midfoot portion 41, particularly on the medial side, is a arch portion 43. Arch shape 44 is the contour shape of the medial side (also called the inner side of the foot) from the forefoot portion 40 to the midfoot portion 41, and mainly refers to the contour shape of the medial side of the arch portion 43. Arch shape 44 is the shape of a part that is usually concave. Outer shape 45 in the vicinity of the ball of the little toe is the contour shape of the lateral side (also called the lateral side of the foot) from the forefoot portion 40 to the midfoot portion 41, and mainly refers to the contour shape of the lateral side of the forefoot portion 40. Outer shape 45 in the vicinity of the ball of the little toe is the shape of a part that usually protrudes.

[0037] The foot length calculation unit 62 calculates the distance between the upper side 22 passing through the toe side end point 24 and the lower side 23 passing through the heel side end point 25 in the bounding box 21, i.e., the length of the left side 26 passing through the inner end point 28, as the foot length 31. The inner end point 28 is located at a predetermined percentage of the left side 26, for example, 72% of the foot length 31 or 28% from the upper side 22. The outer end point 29 is located at a predetermined percentage of the right side 27, for example, 63% of the foot length 31 or 37% from the upper side 22. The foot width calculation unit 64 calculates the length of a straight line connecting the inner end point 28 and the outer end point 29 as the foot width 32.

[0038] FIG. 4 shows a method for measuring the medial indentation amount and the lateral protrusion amount. The medial indentation amount 38 calculated by the indentation amount calculation unit 66 is the width distance between a virtual line passing through the most protruding position in the arch shape 44 and a virtual line passing through the most recessed position. In other words, the medial indentation amount 38 is the distance between the left side 26 passing through the medial end point 28 and the medial reference line 36 passing through the most recessed point 33, which is the most recessed position in the arch shape 44. The medial reference line 36 is a virtual line parallel to the left side 26. The medial indentation amount 38 takes a negative value when there is a recess, and takes a positive value when there is no recess and there is a protruding shape. Here, the virtual line passing through the most recessed point 33 and perpendicular to the medial reference line 36 is defined as the foot width direction line 35. The intersection of the contour line on the lateral side of the midfoot part 41 and the foot width direction line 35 is defined as the lateral reference point 34. An imaginary line passing through the outer reference point 34 and parallel to the right side 27 is defined as an outer reference line 37 .

[0039] The outward protrusion amount 39 calculated by the protrusion amount calculation unit 68 is the width distance between an imaginary line passing through the most protruding point in the contour shape 45 near the ball of the little toe and an imaginary line passing through a specified reference point. In other words, the outward protrusion amount 39 is the distance between the right side 27 passing through the outer end point 29 and the outer reference line 37 passing through the outer reference point 34. The outward protrusion amount 39 is a positive value if there is a protrusion, and is a negative value if there is a recess without protrusion.

[0040] FIG. 5 shows the relationship between pronation type and shoe type. The pronation type 120 stored in the pronation type storage unit 84 includes "neutral pronation" in which the heel is moderately inwardly tilted when landing, i.e., the valgus angle is within the threshold value X, "overpronation" in which the heel is greatly inwardly tilted and the valgus angle exceeds the threshold value X, and "underpronation" in which the heel is little inwardly tilted or outwardly tilted. The heel is also called "supination", but here it is classified as being included in "underpronation" which is little inwardly tilted. Using "neutral pronation" as a standard, the degree of pronation of the foot higher than "neutral pronation" is classified as a first type 110, and the degree of pronation of the foot lower than "neutral pronation" is classified as a second type 111.

[0041] As the shoe type 121, a "pronation compatible model" corresponding to the first type 110 and a "cushion model" corresponding to the second type 111 are stored in the shoe type storage unit 86 as shoe manufacturing models. The "pronation compatible model" is also called a "stability model." These models may have a harder member disposed on the inside of the heel to suppress pronation, may have a mechanism for holding the arch, or may have a function for suppressing inward twisting of the midfoot.

[0042] Figure 6 shows the results of comparing the classification results of subjects using decision tree analysis with the classification results of pronation type based on three-dimensional valgus angle. In the first analysis, decision tree analysis was performed based on all four elements: foot length, foot width, amount of medial concavity, and amount of lateral protrusion. In the first analysis, the results matched with the classification results of pronation type based on three-dimensional valgus angle, and the pronation type could be estimated with 100% accuracy. In the second analysis, decision tree analysis was performed based on two elements: amount of medial concavity and amount of lateral protrusion. In the second analysis, the results matched with the classification results of pronation type based on three-dimensional valgus angle, and the pronation type could be estimated with 100% accuracy.

[0043] In the third analysis, a decision tree analysis was performed based on three factors: foot length, medial concavity, and lateral protrusion. In the third analysis, the results were partially inconsistent with the classification results of pronation type based on the three-dimensional valgus angle, with an estimation accuracy of less than 100%. In the fourth analysis, a decision tree analysis was performed based on three factors: foot width, medial concavity, and lateral protrusion. In the fourth analysis, the results were also in agreement with the classification results of pronation type based on the three-dimensional valgus angle, with an estimation accuracy of 100%.

[0044] In the fifth analysis, a decision tree analysis was performed based on three elements: foot length, foot width, and amount of medial depression. In the fifth analysis, the pronation type was estimated with 100% accuracy, consistent with the classification results of pronation type based on the three-dimensional valgus angle. In the sixth analysis, a decision tree analysis was performed based on three elements: foot length, foot width, and amount of lateral depression. In the sixth analysis, the pronation type was estimated with 100% accuracy, consistent with the classification results of pronation type based on the three-dimensional valgus angle.

[0045] In the seventh analysis, a decision tree analysis was performed based on two factors: foot length and medial depression. In the seventh analysis, the results matched the pronation type classification results based on the three-dimensional valgus angle, and the pronation type could be estimated with 100% accuracy. In the eighth analysis, a decision tree analysis was performed based on two factors: foot length and lateral projection. In the eighth analysis, the results partially did not match the pronation type classification results based on the three-dimensional valgus angle, and the estimation accuracy was less than 100%.

[0046] In the ninth analysis, a decision tree analysis was performed based on two factors: foot width and amount of medial depression. In the ninth analysis, the results matched the pronation type classification results based on the three-dimensional valgus angle, and the pronation type could be estimated with 100% accuracy. In the tenth analysis, a decision tree analysis was performed based on two factors: foot width and amount of lateral projection. In the tenth analysis, the results partially did not match the pronation type classification results based on the three-dimensional valgus angle, and the estimation accuracy was less than 100%.

[0047] According to the above experiments, it is found that the pronation type can be estimated with high accuracy when the decision tree analysis is performed using a combination of elements including the foot width and the amount of medial indentation. On the other hand, all of the decision tree analyses in which the estimation accuracy was less than 100% were performed using a combination of elements including the amount of lateral projection. Therefore, as a suitable combination, the present embodiment employs a foot pronation estimation algorithm based on three elements, (1) foot length, foot width, and amount of medial indentation, or (2) foot width, amount of medial indentation, and amount of lateral projection, which include at least the foot width and the amount of medial indentation. However, in a modified example, a foot pronation estimation algorithm based on a combination of two elements, foot width and amount of medial indentation, is also effective, and a foot pronation estimation algorithm based on a combination of three elements, foot length, foot width, and amount of lateral projection, is also effective. Furthermore, a foot pronation estimation algorithm based on all four elements, foot length, foot width, amount of medial indentation, and amount of lateral projection, is also effective.

[0048] Fig. 7 is a tree diagram that shows a model of the result of decision tree analysis of foot pronation estimation based on three elements: foot length, foot width, and medial concavity. In the example of Fig. 7, measurements with a sample number n of 11 are classified by decision tree analysis with a maximum node depth set to 5. In the root node N10, "foot length" is regressively determined as an explanatory variable and "predetermined value a" is determined as a threshold, and whether or not the foot length is equal to or less than a becomes the branching condition.

[0049] When the foot length is equal to or less than a, leaf node N11, where the squared error is 0 and the number of samples n is 1, is classified as under-pronation based on the valgus angle. When the foot length exceeds a, the branch destination is node N12. In node N12, where the number of samples n is 10, "foot width" is regressively determined as the explanatory variable and "predetermined value b" is determined as the threshold, and the branching condition is whether the foot width is equal to or less than b.

[0050] If the foot width is b or less, the branch destination is node N13. In node N13, where the number of samples n is 2, "foot width" is regressively determined as the explanatory variable and "predetermined value c" is determined as the threshold, and the branch condition is whether the foot width is c or less. If the foot width is c or less, leaf node N14, where the square error is 0 and the number of samples n is 1, is classified as overpronation because its valgus angle exceeds the threshold X. If the foot width exceeds c, leaf node N15, where the square error is 0 and the number of samples n is 1, is also classified as overpronation because its valgus angle exceeds the threshold X.

[0051] If the foot width exceeds b at node N12, the branch destination is node N16. In node N16, where the number of samples n is 8, the explanatory variable is "foot width" and the threshold value is "predetermined value d", and the branch condition is whether the foot width is equal to or less than d. If the foot width is equal to or less than d, the branch destination is node N17. In node N17, where the number of samples n is 3, the explanatory variable is "medial indentation amount" and the threshold value is "predetermined value e", and the branch condition is whether the medial indentation amount is equal to or less than e. If the medial indentation amount is equal to or less than e, leaf node N18, where the square error is 0 and the number of samples n is 2, is classified as neutral pronation based on the average of its valgus angle. If the medial indentation amount exceeds e, leaf node N19, where the square error is 0 and the number of samples n is 1, is also classified as neutral pronation based on its valgus angle.

[0052] If the foot width exceeds d at node N16, the branch destination is node N20. At node N20, where the number of samples n is 5, the "amount of medial indentation" is regressively determined as the explanatory variable and the "predetermined value f" is determined as the threshold, and the branch condition is whether the amount of medial indentation is f or less. If the amount of medial indentation is f or less, leaf node N21, where the square error is about 0.9 and the number of samples n is 4, is classified as overpronation because its average valgus angle exceeds the threshold X. If the amount of medial indentation exceeds f, leaf node N22, where the square error is 0 and the number of samples n is 1, is classified as neutral pronation because of its valgus angle.

[0053] By using decision tree analysis for classification, even if a sample with a significantly off-center valgus angle value such as leaf node N11 is included, the feature amount and its threshold value suitable for classification are determined recursively. Therefore, appropriate classification can be achieved without affecting the overall accuracy of foot pronation estimation.

[0054] According to the procedure determined by the decision tree analysis in FIG. 7, the foot pronation estimation algorithm used by the pronation estimation unit 72 is determined. Here, the estimation process by the pronation estimation unit 72 does not determine all of overpronation, neutral pronation, and underpronation, but only estimates whether it is the first type or the second type as shown in FIG. 6. Therefore, the estimation process for distinguishing between neutral pronation and underpronation is unnecessary. Also, the estimation process is unnecessary when all of the branching destinations are classified in the same way, such as node N13 and node N17. As a result, the foot pronation estimation algorithm adopted by the pronation estimation unit 72 is simplified from the procedure by the decision tree analysis in FIG. 7, and is as shown in the following figure. The specific values ​​a, b, d, and f that are thresholds as branching conditions are determined by the decision tree analysis, become part of the foot pronation estimation algorithm, and are stored in the reference storage unit 82 as reference values ​​for each feature amount.

[0055] FIG. 8 is a flowchart showing the process of foot pronation estimation using a foot pronation estimation algorithm based on three elements: foot length, foot width, and medial indentation. In the example of FIG. 8, branching decisions corresponding to nodes N10, N12, N16, and N20 in the decision tree analysis of FIG. 7 are adopted as the foot pronation estimation algorithm. The image acquisition unit 54 acquires a top view image of the foot of the subject 10 (S10), and the feature extraction unit 60 extracts the foot length, foot width, and medial indentation as features of the foot parts (S11). If the foot length is equal to or less than a predetermined value a (Y in S12), the pronation estimation unit 72 estimates that the foot is of the second type (S17). If the foot length exceeds the predetermined value a (N in S12) and the foot width is equal to or less than a predetermined value b (Y in S13), the pronation estimation unit 72 estimates that the foot is of the first type (S16). If the foot width exceeds a predetermined value b in S13 (N in S13) and is equal to or less than a predetermined value d (Y in S14), it is estimated to be the second type (S17). If the foot width exceeds a predetermined value d in S14 (N in S14) and is equal to or less than a predetermined value f (Y in S15), it is estimated to be the first type (S16), and if the medial concave amount exceeds a predetermined value f (N in S15), it is estimated to be the second type (S17). Based on the result of estimation (S16, S17) of whether the type corresponds to the first type or the second type, the output unit 90 outputs the pronation type (S18), and the foot pronation estimation process ends. In the above flow, the processes of S12, S13, S14, and S15 are branch judgment processes corresponding to nodes N10, N12, N16, and N20 in FIG. 7.

[0056] Fig. 9 is a tree diagram that shows a model of the result of decision tree analysis of foot pronation estimation based on three elements: foot width, medial indentation, and lateral protrusion. In the example of Fig. 9, the measured values ​​with a sample number n of 11 are classified by decision tree analysis with a maximum node depth of 5. In the root node N30, "foot width" is regressively determined as an explanatory variable and "predetermined value b" is determined as a threshold, and whether or not the foot width is equal to or less than b is the branching condition.

[0057] When the foot width is equal to or less than b, the branching destination is node N31. In node N31, where the number of samples n is 2, "foot width" is regressively determined as the explanatory variable and "predetermined value c" is determined as the threshold, and the branching condition is whether the foot width is equal to or less than c. When the foot width is equal to or less than c, leaf node N32, where the square error is 0 and the number of samples n is 1, is classified as overpronation because its valgus angle exceeds the threshold X. When the foot width exceeds c, leaf node N33, where the square error is 0 and the number of samples n is 1, is also classified as overpronation because its valgus angle exceeds the threshold X.

[0058] When the foot width exceeds b at the root node N30, the branch destination is node N34. At node N34, where the number of samples n is 9, the "amount of inner depression" is regressively determined as the explanatory variable and the "predetermined value g" is determined as the threshold, and the branch condition is whether the amount of inner depression is equal to or less than g.

[0059] If the amount of medial indentation is g or less, the branch destination is node N35. In node N35, where the number of samples n is 2, the "amount of medial indentation" is regressively determined as the explanatory variable and the "predetermined value h" is determined as the threshold, and the branch condition is whether the amount of medial indentation is h or less. If the amount of medial indentation is h or less, leaf node N36, where the square error is 0 and the number of samples n is 1, is classified as neutral pronation based on its valgus angle. If the foot width exceeds h, leaf node N37, where the square error is 0 and the number of samples n is 1, is classified as underpronation based on its valgus angle.

[0060] If the amount of medial depression exceeds g at node N34, the branch destination is node N38. In node N38, where the number of samples n is 7, the explanatory variable is "foot width" and the threshold value is "predetermined value i", and the branch condition is whether the foot width is i or less. If the foot width is i or less, the branch destination is node N39. In node N39, where the number of samples n is 3, the explanatory variable is "lateral overhang" and the threshold value is "predetermined value j", and the branch condition is whether the amount of lateral overhang is j or less. If the amount of lateral overhang is j or less, leaf node N40, where the square error is 0 and the number of samples n is 1, is classified as neutral pronation based on its valgus angle. If the amount of lateral overhang exceeds j, leaf node N41, where the square error is 0 and the number of samples n is 1, is also classified as neutral pronation based on its valgus angle.

[0061] If the foot width exceeds i at node N38, the branch destination is node N42. At node N42, where the number of samples n is 5, "foot width" is regressively determined as the explanatory variable and "predetermined value k" is determined as the threshold, and the branch condition is whether the foot width is k or less. When the foot width is k or less, leaf node N43, where the square error is about 0.9 and the number of samples n is 4, is classified as overpronation because its average valgus angle exceeds the threshold X. When the foot width exceeds k, leaf node N44, where the square error is 0 and the number of samples n is 1, is classified as neutral pronation based on its valgus angle.

[0062] According to the procedure determined by the decision tree analysis in FIG. 9, the foot pronation estimation algorithm used by the pronation estimation unit 72 is determined. Here, the estimation process by the pronation estimation unit 72 does not determine all of overpronation, neutral pronation, and underpronation, but only estimates whether it is the first type or the second type as shown in FIG. 6. Therefore, the estimation process that distinguishes between neutral pronation and underpronation is unnecessary. Also, the estimation process is unnecessary when all of the branching destinations are the same classification, such as node N13 and node N17. As a result, the foot pronation estimation algorithm adopted by the pronation estimation unit 72 is simplified from the procedure by the decision tree analysis in FIG. 9, as shown in the following figure. Also, the determination result of node N39, which is the branching process based on the amount of lateral protrusion, is neutral pronation in both cases, and is omitted in the foot pronation estimation algorithm in the following figure. Therefore, the amount of lateral protrusion is not referred to in the foot pronation estimation process, and is substantially equivalent to the foot pronation estimation process based on two elements, the foot width and the amount of medial concavity. The specific values ​​b, g, i, and k that serve as thresholds for branching conditions are determined by decision tree analysis, become part of the foot pronation estimation algorithm, and are stored in the reference memory unit 82 as reference values ​​for each feature.

[0063] FIG. 10 is a flowchart showing the process of foot pronation estimation using a foot pronation estimation algorithm based on three elements, foot width, medial indentation amount, and lateral protrusion amount. In the example of FIG. 10, branching decisions corresponding to nodes N30, N34, N38, and N42 in the decision tree analysis of FIG. 9 are adopted as the foot pronation estimation algorithm. The image acquisition unit 54 acquires a top view image of the foot of the subject 10 (S20), and the feature extraction unit 60 extracts the foot width, medial indentation amount, and lateral protrusion amount as the feature amounts of the foot part (S21). In a modified example, the feature extraction unit 60 may extract only the foot width and medial indentation amount as the feature amounts of the foot part. If the foot length is equal to or less than a predetermined value b (Y in S22), the pronation estimation unit 72 estimates that it is the first type (S26). If the foot length exceeds a predetermined value b (N in S22) and the amount of medial indentation is equal to or less than a predetermined value g (Y in S23), it is estimated to be the second type (S27). If the amount of medial indentation exceeds a predetermined value g in S23 (N in S23) and the foot width is equal to or less than a predetermined value i (Y in S24), it is estimated to be the second type (S27). If the foot width exceeds a predetermined value i in S24 (N in S24) and the foot width is equal to or less than a predetermined value k (Y in S25), it is estimated to be the first type (S26), and if the foot width exceeds a predetermined value k (N in S25), it is estimated to be the second type (S27). Based on the result of estimation of whether the foot corresponds to the first type or the second type (S26, S27), the output unit 90 outputs the pronation type (S28), and the foot pronation estimation process ends. In the above flow, the processes of S22, S23, S24, and S25 are branch determination processes corresponding to the nodes N30, N34, N38, and N42 in FIG.

[0064] The embodiment has been described above. This embodiment is merely an example, and it will be understood by those skilled in the art that various modifications are possible in the combination of each component and each treatment process, and that such modifications are also within the scope of the present invention.

[0065] In a modified example, a side view image of the foot may be used as an auxiliary in addition to the top view image of the foot, and some of the feature amounts of the multiple foot parts may be extracted from the side view image. In another modified example, the pronation estimation unit 72 may acquire information regarding the medical history of the subject 10, and may refer to the information regarding the medical history as an auxiliary in the process of estimating the pronation type.

[0066] In another modified example, a manufacturing ratio of shoe product models corresponding to a first type and shoe product models corresponding to a second type may be adjusted according to the distribution of the pronation types based on the estimation results of the pronation types for a large number of subjects 10. In another modified example, a training method for the posterior tibial muscle, the trunk, and the buttocks may be suggested according to the estimation results of the pronation type for the subjects 10. In another modified example, not only a shoe product model according to the estimation results of the pronation type, but also an insole design according to the estimation results of the pronation type may be suggested.

[0067] Moreover, the above-described embodiment can be generalized to obtain the following aspects. (1) A foot pronation estimation device according to one aspect of the present disclosure, an image acquisition unit that acquires at least a top view image of the foot as an image showing a contour shape of the subject's foot when viewed from above; a feature extraction unit that extracts feature amounts of a predetermined portion indicating a top view contour shape of the foot from the top view image; a pronation estimation unit that estimates a degree of pronation of the foot based on a comparison between the feature amount of the predetermined part and a reference value; Equipped with. The foot pronation estimation device of (1) above provides a measurement subject with a technique for easily determining the pronation type.

[0068] (2) In the foot pronation estimation device according to (1), the feature amount extraction unit extracts feature amounts of a plurality of parts including at least a foot width length and a concave amount of an arch shape from the top view image; The pronation estimation unit may estimate a degree of pronation of the foot based on a comparison between feature amounts of a plurality of parts, including at least the foot width length and the concavity amount of the arch shape, and respective reference values.

[0069] (3) In the foot pronation estimation device according to (2), The feature amounts of the multiple body parts may further include a foot length.

[0070] (4) In the foot pronation estimation device according to (2), The feature amounts of the multiple parts may further include a protrusion amount in the contour shape in the vicinity of the ball of the little toe.

[0071] (5) A foot pronation estimation device according to any one of (1) to (4), The method further includes a storage unit that stores a plurality of pronation types including at least a first type in which the degree of pronation is higher than a predetermined standard and a second type in which the degree of pronation is equal to or lower than a predetermined standard, as types classified according to the degree of pronation of the foot; The pronation estimation unit may estimate which of the plurality of pronation types the predetermined part corresponds to based on a comparison between the feature amount of the predetermined part and a reference value.

[0072] (6) A foot pronation estimation device according to any one of (1) to (5), a storage unit that stores a plurality of shoe types classified according to how the shoe is sized to deal with the pronation of the foot, the shoe types including at least a first shoe type for dealing with a case where the degree of pronation is higher than a predetermined standard, and a second shoe type for dealing with a case where the degree of pronation is equal to or lower than the predetermined standard; a type determining unit that determines a shoe type to be recommended to the subject from among the plurality of shoe types based on the degree of pronation estimated by the pronation estimating unit; The device may further include:

[0073] (7) A foot pronation estimation device according to any one of (1) to (6), A storage unit for storing respective reference values ​​related to the feature amounts of the plurality of parts, The pronation estimation unit may estimate a degree of pronation based on a comparison between the extracted feature amount and the stored reference value.

[0074] (8) A foot pronation estimation device according to any one of (1) to (7), The image acquisition section may acquire, as the top view image, an image showing a top view contour shape of the subject's foot in a sitting position.

[0075] (9) A foot pronation estimation device according to any one of (1) to (8), The image acquisition unit further acquires a side view image of the subject's foot, The feature amount extracting section may extract the feature amount of the predetermined portion from at least one of the top view image and the side view image.

[0076] (10) A method for estimating foot pronation according to one aspect of the present disclosure, acquiring at least a top view image of the foot as an image showing a contour shape of the subject's foot when viewed from above; extracting feature amounts of predetermined portions indicating a top view contour shape of the foot from the top view image by image processing; a step of estimating a degree of pronation of the foot based on a comparison process between the feature amount of the predetermined part and a reference value; Equipped with.

[0077] (11) A foot pronation estimation program according to an embodiment of the present disclosure, A function of acquiring at least a top view image of the foot as an image showing a contour shape of the subject's foot when viewed from above; A function of extracting feature amounts of a predetermined portion indicating a top view contour shape of a foot from the top view image; a function of estimating a degree of pronation of the foot based on a comparison between the feature amount of the predetermined part and a reference value; A function of outputting information indicating a result of the estimation; is demonstrated on a computer. [Explanation of symbols]

[0078] 10 Measurement subject, 30 Top view image, 31 Foot length, 32 Foot width, 38 Medial concavity amount, 39 Lateral protrusion amount, 44 Arch shape, 45 Outer shape of area near little toe ball, 50 Foot pronation estimation device, 54 Image acquisition unit, 60 Feature extraction unit, 72 Pronation estimation unit, 74 Type determination unit, 80 Memory unit, 110 First type, 111 Second type, 120 Pronation type, 121 Shoe type.

Claims

1. an image acquisition unit that acquires at least a top view image of the foot as an image showing a contour shape of the subject's foot when viewed from above; a feature extraction unit that extracts feature amounts of a predetermined portion indicating a top view contour shape of the foot from the top view image; a pronation estimation unit that estimates a degree of pronation of the foot based on a comparison between the feature amount of the predetermined part and a reference value; A foot pronation estimation device comprising:

2. the feature amount extraction unit extracts feature amounts of a plurality of parts including at least a foot width length and a concave amount of an arch shape from the top view image; The foot pronation estimation device according to claim 1, characterized in that the pronation estimation unit estimates the degree of pronation of the foot based on a comparison of feature amounts of multiple parts, including at least the length of the foot width and the amount of concavity of the arch shape, with their respective reference values.

3. The foot pronation estimation device according to claim 2 , wherein the feature amounts of the multiple parts further include a foot length.

4. The foot pronation estimation device according to claim 2 , wherein the feature amounts of the multiple parts further include a protrusion amount in the contour shape in the vicinity of the ball of the little toe.

5. The method further includes a storage unit that stores a plurality of pronation types including at least a first type in which the degree of pronation is higher than a predetermined standard and a second type in which the degree of pronation is equal to or lower than a predetermined standard, as types classified according to the degree of pronation of the foot; The foot pronation estimation device according to claim 1 , wherein the pronation estimation unit estimates which of the plurality of pronation types the specific part corresponds to based on a comparison between the feature amount of the specific part and a reference value.

6. a storage unit that stores a plurality of shoe types classified according to how the shoe is sized to deal with the pronation of the foot, the shoe types including at least a first shoe type for dealing with a case where the degree of pronation is higher than a predetermined standard, and a second shoe type for dealing with a case where the degree of pronation is equal to or lower than the predetermined standard; a type determining unit that determines a shoe type to be recommended to the subject from among the plurality of shoe types based on the degree of pronation estimated by the pronation estimating unit; The foot pronation estimation device according to claim 1, further comprising:

7. A storage unit for storing a reference value related to the feature amount of the predetermined part, 2. The foot pronation estimation device according to claim 1, wherein the pronation estimation unit estimates the degree of pronation based on a comparison between the extracted feature amount and the stored reference value.

8. 2. The foot pronation estimating device according to claim 1, wherein the image acquisition unit acquires, as the top view image, an image showing a top view contour shape of the subject's foot in a sitting position.

9. The image acquisition unit further acquires a side view image of the subject's foot, The foot pronation estimation device according to claim 1 , wherein the feature amount extracting section extracts the feature amount of the predetermined part from at least one of the top view image and the side view image.

10. acquiring at least a top view image of the foot as an image showing a contour shape of the subject's foot when viewed from above; extracting feature amounts of predetermined portions indicating a top view contour shape of the foot from the top view image by image processing; a step of estimating a degree of pronation of the foot based on a comparison process between the feature amount of the predetermined part and a reference value; A foot pronation estimation method comprising:

11. A function of acquiring at least a top view image of the foot as an image showing a contour shape of the subject's foot when viewed from above; A function of extracting feature amounts of a predetermined portion indicating a top view contour shape of a foot from the top view image; a function of estimating a degree of pronation of the foot based on a comparison between the feature amount of the predetermined part and a reference value; A function of outputting information indicating a result of the estimation; A foot pronation estimation program characterized by causing a computer to perform the above.

Images