JP2012198607A - Coordinate correction function generation device, input device, coordinate correction function generation method, coordinate correction method and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To generate a coordinate correction function for reducing errors between a detection position and a real position in realtime.SOLUTION: There is provided a coordinate correction function generation device comprising: a noise removal unit that uses coordinates Xobtained at n-th, coordinates Xobtained at (n-p)-th (p=1 to P) and coordinates Xobtained at (n+q)-th (q=1 to Q) from among a plurality of coordinates obtained by a touch sensor, to calculate coordinates X' in which a noise corresponding to the coordinates Xis removed; and a mapping generation unit that generates mapping function Fwith which the coordinate X' is output when coordinate Xobtained at (n-m)-th (m=1 to M) is input, through machine learning.

Description

  The present technology relates to a coordinate correction function generation device, an input device, a coordinate correction function generation method, a coordinate correction method, and a program.

  In recent years, various types of touch sensors have been developed (see, for example, Patent Document 1 below), and there are many methods already in practical use. Among them, a great deal of attention has been attracted to the capacitive touch sensor from the viewpoint of good operability and high durability. The capacitive touch sensor uses the change in capacitance that occurs between the electrode in the touch sensor and the operating body when the operating body (for example, a finger) approaches the touch sensor. Is detected. The change in capacitance that occurs between the electrode and the operating body occurs even when the operating body does not contact the touch sensor. For this reason, the touch sensor reacts even if the operating body approaches the surface or the operating body touches the surface lightly. From such a good response, the user can obtain a good operational feeling. In addition, the capacitive touch sensor can detect the positions of a plurality of operating bodies that are close to or in contact with the surface thereof.

JP 2010-39515 A

  Capacitive touch sensors have been mounted on various electronic devices for reasons such as good operability. For example, a capacitive touch sensor is mounted on various portable electronic devices such as a mobile phone, a portable information terminal, a portable music player, and a portable game machine. Capacitive touch sensors are also expected to be applied to relatively large electronic devices such as television receivers, personal computers, car navigation systems, digital signage terminals, and ATMs.

  However, the touch position (hereinafter, detected position) detected by the capacitive touch sensor may deviate from the actually touched position (hereinafter, actual position). Therefore, various studies for reducing the error between the detected position and the actual position have been made in various places. Here, the capacitive touch sensor is taken as an example, but methods for reducing the error between the detection position and the actual position have been studied for other various sensors that detect the position. The present inventor has also studied a method for reducing the error between the detected position and the actual position.

  The present technology has been devised in view of the above circumstances, and can generate a coordinate correction function for reducing an error between a detected position and an actual position without causing a processing delay. It is intended to provide a new and improved coordinate correction function generation device, input device, coordinate correction function generation method, coordinate correction method, and program that are possible.

In order to solve the above-described problem, according to an aspect of the present technology, among the plurality of coordinates acquired by the touch sensor, the nth acquired coordinate _Xn and the (n−p) th (p = 1 to _P) and the coordinate X _(n−q) acquired in the (n + q) th (q = 1 to Q) coordinate X _{(n−p). The} noise removing unit for calculating the coordinate X _n ′ from which the noise corresponding to _n has been removed and the coordinate X (n−m) acquired in the (n−m) th (m = 1 to M ₎ are input. There is provided a coordinate correction function generation device comprising: a function generation unit that generates a function Fn that outputs the coordinate X _n ′ by machine learning.

The function generation unit expresses the function F _n by a linear combination of the coordinates X _(n−m) (m = 1 to M), and the coordinates X _n ′ and the coordinates X _(n−m) ( The coupling coefficient may be calculated by the least square method using the actually measured values of m = 1 to M).

The noise removing unit averages the coordinates _Xn , the coordinates X _(np) (p = 1 to P), and the coordinates X _{(n + q)} (q = 1 to Q), and The coordinate X _n ′ may be calculated.

Further, the noise removing unit increases the coordinate X _n and a value obtained by multiplying the coordinate X _(n−p) by a weight D _p (D _p <1) that decreases as p increases, and q increases. The coordinate X _n ′ may be calculated by averaging a value obtained by multiplying the weight X _q (D _q <1) by the weight D _q (D _q <1), the value of which decreases with time, on the coordinate X _{(n + q)} .

The noise removing unit inputs the coordinates _Xn , the coordinates X _(np) (p = 1 to P), and the coordinates X _{(n + q)} (q = 1 to Q) to the Butterworth filter. Thus, the coordinates X _n ′ may be calculated.

In addition, the noise removing unit is a regression that shows a relationship among the coordinates _Xn , the coordinates X _(n−p) (p = 1 to P), and the coordinates X _{(n + q)} (q = _{1 to Q).} calculating a linear or regression curve, the coordinates on the regression line or regression curve corresponding to the coordinates X _n may be configured so as to the coordinate X _{n '.}

In addition, the coordinate correction function generation device classifies the function F _n based on the movement speed calculated by the speed calculation unit, the speed calculation unit that calculates the movement speed at the coordinate X _n , and each class And a function classifying unit that generates a function representative of.

Also, the coordinate correcting function generating device includes a speed calculator for calculating the moving velocity in the coordinate X _n, divided grid of the touch sensor into a plurality of regions, the region determining region including the coordinate X _n A function classifying unit that classifies the function F _n based on the moving speed calculated by the speed calculating unit and the region determined by the region determining unit, and generates a function representing each class; May be further provided.

  The coordinate correction function generation device includes the touch sensor and a coordinate correction unit that inputs the coordinates acquired by the touch sensor to the function generated by the function generation unit and corrects the coordinates. You may further provide the coordinate information acquisition part which acquires the information regarding the coordinate which the user input using the said touch sensor from the input device. In this case, the noise removal unit calculates coordinates from which noise has been removed using a coordinate group indicated by the information acquired by the coordinate information acquisition unit, and the function generation unit is acquired by the coordinate information acquisition unit. A function is generated using the coordinate group indicated by the information and the coordinates calculated by the noise removing unit.

  The coordinate information acquisition unit may be configured to acquire information indicating the user's operation pattern as information on the coordinates. Furthermore, the noise removal unit may be configured to calculate coordinates from which noise has been removed using a coordinate group corresponding to the operation pattern. The function generation unit may be configured to generate a function using a coordinate group corresponding to the operation pattern and the coordinates calculated by the noise removal unit.

In order to solve the above problem, according to another aspect of the present technology, the touch sensor that detects the touched coordinate and the nth acquired among the plurality of coordinates acquired in advance by the touch sensor. Coordinates _Xn , coordinates (np) acquired in the (n−p) th (p = 1 to _P), and (n + q) th (q = 1 to Q). based on the coordinates X _(n-q) and the coordinates X _{n 'corresponding} to the coordinates X _n in which noise is removed using, generated by machine learning, the first (n-m) th (m = 1 To the function F _n that outputs the coordinate X _n ′ when the acquired coordinate X _(n−m) is input, the coordinate detected by the touch sensor is input and the coordinate is corrected. And an input device including a coordinate correction unit.

  The input device includes an operation pattern detection unit that detects an operation pattern of a user using the touch sensor, and an operation detected by the operation pattern detection unit among functions prepared for each predetermined operation pattern. And a function selection unit that selects a function corresponding to the pattern. In this case, the coordinate correction unit inputs the coordinates detected by the touch sensor to the function selected by the function selection unit and corrects the coordinates.

  In addition, the input device includes an operation body size detection unit that detects the size of the operation body that operates the touch sensor, and a function prepared for each size of the operation body by the operation body size detection unit. And a function selecting unit that selects a function corresponding to the detected size of the operating tool. In this case, the coordinate correction unit inputs the coordinates detected by the touch sensor to the function selected by the function selection unit and corrects the coordinates.

  The input device may further include a deviation degree detection unit that detects a degree of deviation between the coordinates detected by the touch sensor and the coordinates corrected by the coordinate correction unit. In this case, the coordinate correction unit changes the function when the deviation degree detected by the deviation degree detection unit is larger than a predetermined threshold value.

In order to solve the above problem, according to another aspect of the present technology, among the plurality of coordinates acquired by the touch sensor, the nth acquired coordinate Xn and the (n−p) th coordinate The coordinates X _(n−p) acquired at (p = 1 to _P) and the coordinates X _(n−q) acquired at the (n + q) th (q = 1 to _Q) are used, and a noise removal step of noise corresponding to the coordinates _{X n} is calculated coordinates _{X n} 'is removed, the first (n-m) th (m = 1 to m) on the acquired coordinate _{X (n-m)} A function generation step of generating a function F _n that outputs the coordinate X _n ′ when input by machine learning is provided.

In order to solve the above-described problem, according to another aspect of the present technology, among the plurality of coordinates acquired by the touch sensor, the nth acquired coordinate _Xn and the (n−p) th coordinate. The coordinates X _(n−p) acquired for the th (p = 1 to _P) and the coordinates X _(n−q) acquired for the (n + q) th (q = 1 to _Q) , A noise removing function for calculating coordinates X _n ′ from which noise corresponding to the coordinates X _n has been removed, and the coordinates X (nm) acquired at the (n−m) th (m = 1 to M _). A program for causing a computer to realize a function generation function that generates a function F _n that outputs the coordinates X _n ′ by machine learning when the input is input is provided.

In order to solve the above problem, according to another aspect of the present technology, a coordinate detection step in which touched coordinates are detected by a touch sensor, and a plurality of coordinates acquired in advance by the touch sensor. , The nth acquired coordinate _Xn , the (n−p) th (p = 1 to P) acquired coordinate X _(n−p) , and the (n + q) th (q = 1 to n) coordinates obtained in _{Q) X (n-q)} , based on the coordinate X _{n 'corresponding} to the coordinates X _n in which noise is removed using, generated by machine learning, the first (n-m The coordinates detected by the touch sensor are input to the function F _n that outputs the coordinates X _n ′ when the coordinates X ₍ nm) acquired in the first (m = 1 to M) are input. And a coordinate correction step for correcting the coordinates. That.

In order to solve the above-mentioned problem, according to another aspect of the present technology, a coordinate detection function for detecting coordinates touched by a touch sensor, and a plurality of coordinates acquired in advance by the touch sensor, The nth acquired coordinate Xn, the (n−p) th (p = 1 to P) acquired coordinate X _(n−p) , and the (n + q) th (q = 1 to Q) based on the coordinates X _{n 'corresponding} to the coordinates X _n in which noise is removed using the obtained coordinates X _(n-q), to have been generated by a machine learning, the first (n-m) th When the coordinates X ₍ nm) acquired at (m = 1 to M) are input, the coordinates detected by the touch sensor are input to the function F _n that outputs the coordinates X _n ′. A program for realizing a coordinate correction function for correcting the coordinates on a computer. Beam is provided.

  In order to solve the above problem, according to another aspect of the present technology, a computer-readable recording medium on which the above-described program is recorded is provided.

  As described above, according to the present technology, it is possible to generate a coordinate correction function for reducing an error between the detected position and the actual position without causing a processing delay.

It is explanatory drawing which shows the electrode structure of an electrostatic capacitance type touch sensor. It is explanatory drawing which shows the detection method of the touch position by an electrostatic capacitance type touch sensor. It is explanatory drawing for demonstrating the detection accuracy of an electrostatic capacitance type touch sensor. It is explanatory drawing for demonstrating the noise reduction method using a moving average. It is explanatory drawing for demonstrating the noise reduction method using a moving average. It is explanatory drawing which shows the structure of the information processing apparatus which concerns on one Embodiment of this technique. It is explanatory drawing for demonstrating the property of the coordinate correction based on a moving average. It is explanatory drawing for demonstrating the characteristic of the correction function which concerns on this embodiment. It is explanatory drawing which shows the flow of the correction process which concerns on this embodiment. It is explanatory drawing which shows the flow of the update process of the correction function which concerns on this embodiment. It is explanatory drawing which shows the flow of the update process of the correction function which concerns on this embodiment. It is explanatory drawing which shows the structure of the information processing system which concerns on this embodiment. It is explanatory drawing for demonstrating the production | generation method of the teacher data which concerns on this embodiment. It is explanatory drawing for demonstrating the production | generation method of the correction function which concerns on this embodiment. It is explanatory drawing for demonstrating the classification method of the correction function which concerns on this embodiment. It is explanatory drawing for demonstrating the classification method of the correction function which concerns on this embodiment. It is explanatory drawing which shows the flow of the production | generation process and classification | category process of a correction function which concern on this embodiment. It is explanatory drawing which shows the flow of the production | generation process and classification | category process of a correction function which concern on this embodiment. It is explanatory drawing which shows the flow of the production | generation process and classification | category process of a correction function which concern on this embodiment. It is explanatory drawing which showed the system configuration example which concerns on the modification of this embodiment. It is explanatory drawing which showed the system configuration example which concerns on the modification of this embodiment. It is explanatory drawing which shows the hardware structural example which can implement | achieve the function of the information processing apparatus which concerns on this embodiment, and an information processing system.

  Hereinafter, preferred embodiments according to the present technology will be described in detail with reference to the accompanying drawings. In addition, in this specification and drawing, about the component which has the substantially same function structure, duplication description is abbreviate | omitted by attaching | subjecting the same code | symbol.

[About the flow of explanation]
Here, the flow of explanation described below will be briefly described.

  First, the electrode structure and its properties of a capacitive touch sensor will be briefly described with reference to FIGS. A sensor other than the capacitive touch sensor will be described. Next, a noise reduction method using a moving average will be described with reference to FIGS. 4 and 5. Next, the configuration and operation of the information processing apparatus according to an embodiment of the present technology will be described with reference to FIGS. Next, the configuration and operation of the information processing system 100 according to the embodiment will be described with reference to FIGS.

  Next, a system configuration capable of realizing the technique according to the embodiment will be briefly described with reference to FIGS. Next, a hardware configuration example capable of realizing the functions of the information processing apparatus according to the embodiment of the present technology will be described with reference to FIG. Finally, the technical idea of the embodiment will be summarized and the effects obtained from the technical idea will be briefly described.

(Description item)
1: Introduction 1-1: Structure of capacitive touch sensor and touch position detection method 1-2: Detection accuracy of capacitive touch sensor 1-3: Other sensors 1-4: Noise reduction method 2 : Embodiment 2-1: Configuration of information processing apparatus 2-2: Operation of information processing apparatus 2-3: Configuration of information processing system 100 2-4: Operation of information processing system 100 2-5: Summary of system configuration
2-5-1: Configuration example 1
2-5-2: Configuration example 2
3: Hardware configuration 4: Summary

<1: Introduction>
First, the structure of the capacitive touch sensor, the detection accuracy of the capacitive touch sensor, the detection accuracy of other sensors, and a noise reduction method will be briefly described.

[1-1: Structure of Capacitive Touch Sensor and Touch Position Detection Method]
First, the electrode structure of the capacitive touch sensor will be described with reference to FIG. In addition, a method for detecting a touch position using a capacitive touch sensor will be described with reference to FIG. FIG. 1 is an explanatory diagram for describing an electrode structure of a capacitive touch sensor. FIG. 2 is an explanatory diagram for explaining a method of detecting a touch position by a capacitive touch sensor. Here, as an example, a diamond type electrode structure will be described as an example.

  As shown in FIG. 1, the capacitive touch sensor includes a plurality of X electrodes wired along the X direction and a plurality of Y electrodes wired along the Y direction. Further, the X electrode and the Y electrode are arranged and wired so that the rectangular portion (electrode pad) of the X electrode and the rectangular portion (electrode pad) of the Y electrode are evenly exposed when viewed in the Z direction. Although only a few electrodes are shown here for simplicity, there are actually a large number of electrodes.

  As shown in FIG. 2, when a dielectric such as a finger (hereinafter referred to as an operating body) approaches the X electrode (Y electrode), electrostatic coupling is formed between the operating body and the X electrode (Y electrode), and X The capacitance of the electrode (Y electrode) increases. Therefore, it is possible to detect the proximity of the operating body by detecting a change in capacitance. Further, by monitoring the capacitance of each X electrode and each Y electrode, it is possible to detect the position where the operating body is in contact with or close to.

[1-2: Detection accuracy of capacitive touch sensor]
Next, the detection accuracy of the capacitive touch sensor will be described with reference to FIG. As shown in the left diagram of FIG. 3, when the operator touches the touch surface in an oblique direction and linearly traces, the locus of the touch position as shown in the right diagram of FIG. 3 is detected. As is apparent from the right diagram of FIG. 3, the locus of the touch position detected by the capacitive touch sensor is periodically wavy. That is, the trajectory drawn by the user with the operating tool does not match the trajectory actually detected by the capacitive touch sensor. In other words, there is an error between the coordinates of the position where the operating body approaches or touches the touch surface and the coordinates of the actually detected touch position.

  Thus, if a coordinate different from the position where the operating tool is close or in contact is detected, a command not intended by the user may be executed. In addition, if the locus of the touch position detected during the drag operation undulates, the object that moves following the drag operation may move unnaturally or the gesture may not be recognized correctly. In addition, when the user performs an operation in which the direction of movement, such as scrolling the screen, is performed, the operation speed may fluctuate, which may make the user feel uncomfortable. In order to reduce the above error, for example, the interval between the X electrode and the Y electrode may be narrowed.

  However, when the number of electrodes increases, the time required for scanning the touch position increases, and the response decreases. Furthermore, the manufacturing cost increases due to the increase in the number of terminals. For these reasons, attention has been paid to a technique for reducing the above error while maintaining the size of the grid formed by the X electrode and the Y electrode.

[1-3: Other sensors]
By the way, not only the capacitive touch sensor having the diamond-shaped electrode structure described above, but also in the capacitive touch sensor having another electrode structure and the touch sensor adopting a method other than the capacitive type, A technique for improving the detection accuracy of the touch position is required. In addition to the touch sensor, in the field of an optical sensor that detects a position using infrared rays, visible light, or the like, it is required to realize a technique for accurately detecting the position of a light source. The detection results by these sensors include the influence of noise caused by various factors. For this reason, various researches have been conducted on various techniques for removing the influence of noise contained in the detection results of sensors that detect positions by electrical or optical methods.

[1-4: Noise reduction method]
As a method of reducing noise included in the detection result of the sensor, there is a method of averaging the measurement values measured by the sensor and suppressing variation in the measurement values. Here, a noise reduction method using a moving average will be described with reference to FIGS. 4 and 5. 4 and 5 are explanatory diagrams for explaining a noise reduction method using a moving average.

Assume that the measured values X _{1 to} X ₈ are measured by the sensor at times t ₁ to t ₈ . When the measurement values X _{1 to} X ₈ are plotted in the coordinate space with the time t and the measurement value X as coordinate axes, the result is as shown in FIG. First, consider the processing to reduce noise measurements X _3. When to be considered the moving average of five points, the processing for reducing noise measurements X ₃ are each 2-point around the measured value X ₃ are extracted, the average value X of the five points, including the measured value X _{3 3} 'is calculated. That is, the average value X ₃ ′ of the measured values X _{1 to} X ₅ is calculated. A point corresponding to the average value X ₃ ′ will be expressed as P (t ₃ ).

When the point P (t ₃ ) is obtained, next, a process for reducing the noise of the measurement value X ₄ is executed. In the process of reducing the noise measurements X ₄ are each before and after the measurement value X ₄ are extracted, the mean value X ₄ of 5 points including the measured values X _{4 'is} calculated. That is, as shown in FIG. 5, the range of measurement values used for calculating the average value X ₄ ′ moves (A (t ₃ ) → A (t ₄ ) as the measurement values targeted for noise reduction processing move. )). Then, an average value X ₄ ′ of the measured values X _{2 to} X ₆ is calculated. In this way, the average value included in the range is sequentially calculated while moving the range of the measurement value. The average values X ₃ ′, X ₄ ′,... Calculated using the moving average are reduced values included in the measured values X ₃ , X ₄ ,.

However, calculated as shown in FIG. 5, _'for using the measured values X ₁ to X ₅ for calculating the average value X ₃ to wait at least until the time t _5' average X ₃ measurements X ₃ a Can not do it. Therefore, a delay Δt = t ₅ −t ₃ occurs. This delay Δt occurs in the same manner when calculating the average values X ₄ ′, X ₅ ′,. That is, when a noise reduction method using a moving average is applied, a delay Δt occurs between the position detection by the sensor and the output of the position information after noise reduction. The occurrence of such a delay Δt causes a decrease in response and gives the user a feeling of strangeness. Therefore, there is a demand for a technique for removing noise included in the detection result by the sensor while suppressing the occurrence of the delay Δt.

<2: Embodiment>
Hereinafter, an embodiment of the present technology will be described.

[2-1: Configuration of information processing apparatus]
First, the configuration of the information processing apparatus according to the present embodiment will be described with reference to FIG. FIG. 6 is an explanatory diagram for explaining the configuration of the information processing apparatus according to the present embodiment.

(Overall configuration)
As illustrated in FIG. 6, the information processing apparatus according to the present embodiment is mainly configured by an input device 10, an information processing unit 20, and a display unit 30. The input device 10 mainly includes a touch sensor 11, a coordinate storage unit 12, a coordinate correction unit 13, a function storage unit 14, and a function update unit 15. Instead of the touch sensor 11, it is also possible to use a type of sensor that detects a position indicated by the user using light such as infrared rays or visible light. However, in the following, the description will be made with the acquisition of the coordinates of the touch position from the touch sensor 11 as an example.

  When the user brings an operating body such as a finger close to or in contact with the touch sensor 11, the touch sensor 11 detects the position of the operating body that has approached or touched (hereinafter referred to as a touch position). The coordinates indicating the touch position detected by the touch sensor 11 are stored in the coordinate storage unit 12. The coordinates stored in the coordinate storage unit 12 are read out by the coordinate correction unit 13 and the function update unit 15 as necessary. The coordinate correction unit 13 corrects the latest coordinates stored in the coordinate storage unit 12 using a function stored in advance in the function storage unit 14 (hereinafter, correction function). The coordinates corrected by the coordinate correction unit 13 are referred to as corrected coordinates.

  The correction coordinates are output from the coordinate correction unit 13 and input to the information processing unit 20. When the correction coordinates are input, the information processing unit 20 executes a predetermined process based on the input correction coordinates. Then, the information processing unit 20 displays the executed processing result on the display unit 30. In addition, the function update unit 15 updates the correction function stored in the function storage unit 14. For example, the function update unit 15 acquires a correction function suitable for the user's operation pattern, the size of the operation tool, and the like from the information processing system 100, and updates the correction function stored in the function storage unit 14 with the acquired correction function. .

(Detailed configuration of the coordinate correction unit 13)
Here, the configuration of the coordinate correction unit 13 will be described in more detail.

  As described above, the coordinate correction unit 13 corrects the latest coordinates (hereinafter, target coordinates) stored in the coordinate storage unit 12 using the correction function stored in the function storage unit 14. As a coordinate correction method, for example, correction by moving average, correction by weighted average, correction by nonlinear fitting, correction by Butterworth filter, and the like are conceivable. As shown in FIG. 7, the correction by moving average is a method of averaging the target coordinates and a predetermined number of coordinates acquired before and after the target coordinates and setting the average value as the correction coordinates.

  The correction by the weighted average is a method in which a predetermined number of coordinates acquired before and after the target coordinates are respectively weighted, and an average value of the weighted coordinates is set as the correction coordinates. The weighting used in this method is performed using, for example, a weight value less than 1 such that the value decreases as the acquisition time moves away from the acquisition time of the target coordinates. Further, the correction by nonlinear fitting is a method of fitting a coordinate group with a nonlinear function and setting the value of the nonlinear function at the acquisition time of the target coordinates as the corrected coordinates. The correction by the Butterworth filter is a method in which the target coordinates and a predetermined number of coordinates acquired before and after the target coordinates are input to the Butterworth filter, and the output value of the Butterworth filter is set as the correction coordinates.

All the above methods use coordinates acquired after the target coordinates. For example, as shown in FIG. 7, when a moving average using five coordinates is considered, when the correction coordinates X ′ (t ₃ ) of the target coordinates X (t ₃ ) acquired at time t ₃ are obtained, the target coordinates In addition to X (t ₃ ), coordinates X (t ₁ ), X (t ₂ ), X (t ₄ ), and X (t ₅ ) acquired before and after time t ₃ are required. For this reason, the corrected coordinate X ′ (t ₃ ) cannot be calculated until the coordinate X (t ₅ ) is acquired at time t ₅ . Therefore, when a correction method as shown in FIG. 7 is used, a processing delay of at least Δt = t ₅ −t ₃ occurs when calculating the correction coordinate X ′ (t ₃ ). Note that processing delay occurs for the same reason when the other correction methods described above are used.

Therefore, the coordinate correction unit 13 corrects the target coordinates using a correction function F having characteristics as shown in FIG. The correction function F is stored in advance in the function storage unit 14. As shown in FIG. 8, the correction function F has a characteristic of outputting correction coordinates when the target coordinates and a predetermined number of coordinates acquired before the target coordinates are input. For example, if you enter the coordinates _X (t 1) _~X a _{(t 5)} the correction function F, the correction coordinates X _{'(t 5)} corresponding to the coordinates X _{(t 5)} is output. That is, the coordinate correction unit 13 uses the coordinates X (t ₁ ) to X (t ₄ ) already stored in the coordinate storage unit 12 at the time when the target coordinate X (t ₅ ) is acquired, and uses the corrected coordinates X ′. (T ₅ ) can be obtained. Therefore, the processing delay Δt described above does not occur.

  The correction function F is generated by machine learning. Therefore, the correction function F based on an arbitrary correction method can be generated. A method for generating the correction function F will be described later.

  The detailed configuration of the coordinate correction unit 13 has been described above.

(Detailed configuration of the function update unit 15)
Next, the configuration of the function update unit 15 will be described in more detail.

  As described above, the function update unit 15 updates the correction function stored in the function storage unit 14. As described above, the correction function stored in the function storage unit 14 is generated by machine learning. Therefore, when the student data used when performing machine learning does not match the user's operation pattern or the characteristics of the operating body (hereinafter referred to as user characteristics), it is difficult to correct correctly using the correction function. Therefore, the function updating unit 15 acquires a correction function that matches the user characteristics from the information processing system 100, and updates the correction function stored in the function storage unit 14 with the acquired correction function.

  For example, the magnitude of noise to be removed by correction differs between when operating with the fingertip and when operating with the flat of the finger. Further, the magnitude of noise to be removed by correction differs between when the operating body is moved quickly and when it is moved slowly. Furthermore, the magnitude of noise to be removed by correction differs between when the touch sensor 11 is operated while being lightly pressed and when it is operated while being pressed strongly. Further, the magnitude of noise to be removed by the correction differs depending on whether the size of the operating tool is large or small. Therefore, the correction function that can effectively remove noise differs depending on the user characteristics.

  In order to generate a correction function that matches the user feature, a coordinate group reflecting the user feature or information indicating the user feature is required. Therefore, the function update unit 15 transmits information indicating the coordinate group or user characteristics stored in the coordinate storage unit 12 (hereinafter referred to as user feedback) to the information processing system 100. In the information processing system 100, a correction function is newly generated according to the user feedback, or a correction function that matches the user feedback is selected from correction functions prepared in advance for each user feature. The function update unit 15 receives the correction function generated or selected by the information processing system 100 based on the transmitted user feedback, and updates the correction function stored in the function storage unit 14 with the correction function.

  The function updating unit 15 may be configured to update the correction function at a predetermined cycle, or may be configured to update the correction function when the accuracy of correction using the correction function is low. Also good. The accuracy of the correction by the correction function can be evaluated using the sum of squares of errors between the correction coordinates calculated by the correction function and the target coordinates, an average error, or the like. For example, the function updating unit 15 obtains a square sum of errors regarding a predetermined number of coordinates actually detected, and updates the correction function when the square sum exceeds a predetermined threshold. At this time, the function update unit 15 obtains the sum of squares of errors using the correction function stored in the function storage unit 14. Further, the function update unit 15 may be configured to evaluate the accuracy of correction using an error between teacher data obtained in a situation where the user is actually using the target coordinate.

  Note that the function update unit 15 acquires user feature information that matches the correction function stored in the function storage unit 14, and the user feature indicated in the acquired information is different from the actually detected user feature. The correction function may be updated. In this case, the function update unit 15 has a function of detecting a user feature. For example, the function update unit 15 has a function of calculating an average moving speed of the operating tool using the coordinates stored in the coordinate storage unit 12. The function updating unit 15 has a function of acquiring information such as the size, pressing force, and pressing area of the operating tool that can be detected by the touch sensor 11 and calculating an average value of values indicated by the acquired information. Note that the function update unit 15 may be configured to calculate another statistic such as a median value instead of the average value as information indicating the user characteristics.

  The detailed configuration of the function update unit 15 has been described above.

  As described above, the coordinate correction unit 13 corrects the target coordinates using the coordinates acquired before the target coordinates. Therefore, it is possible to correct in real time coordinate errors caused by the influence of device characteristics of the touch sensor 11 and external noise. In addition, by updating the correction function and using a correction function that matches the user characteristics, it is possible to remove noise with high accuracy.

  The configuration of the information processing apparatus according to the present embodiment has been described above.

[2-2: Operation of information processing apparatus]
Next, the operation of the information processing apparatus according to the present embodiment will be described with reference to FIGS. 9 to 11 are explanatory diagrams for explaining the operation of the information processing apparatus according to the present embodiment.

(About correction processing)
First, FIG. 9 will be referred to. The process shown in FIG. 9 is mainly executed by the coordinate correction unit 13. As shown in FIG. 9, first, the coordinate correction unit 13 acquires the latest coordinates from the coordinate storage unit 12 as target coordinates (S101). Next, the coordinate correction unit 13 acquires past coordinates detected before the target coordinates from the coordinate storage unit 12 (S102). Next, the coordinate correction unit 13 inputs the latest and past coordinates acquired in steps S101 and S102 to the correction function (S103). Next, the coordinate correction unit 13 outputs the output value of the correction function as the correction coordinates (S104), and ends a series of processes related to the correction process.

(About correction function update processing)
Reference is now made to FIG. The process shown in FIG. 10 is mainly executed by the function update unit 15. As illustrated in FIG. 10, first, the function update unit 15 acquires a coordinate group from the coordinate storage unit 12 (S111). Next, the function updating unit 15 inputs the coordinate group acquired in step S111 to the correction function (S112). Next, the function updating unit 15 compares the correction coordinates output from the correction function with the target coordinates (S113). For example, the function updating unit 15 obtains a sum of squares of errors between the corrected coordinates and the target coordinates, an average error, and the like, and evaluates a deviation width between the two coordinates. Next, the function updating unit 15 determines whether or not the deviation width of the coordinate value is greater than or equal to a predetermined value (S114).

  If the deviation width of the coordinate value is greater than or equal to the predetermined value, the function updater 15 advances the process to step S115. On the other hand, when the deviation width of the coordinate values is less than the predetermined value, the function update unit 15 ends a series of processes related to the update of the correction function. When the process proceeds to step S115, the function update unit 15 updates the correction function (S115), and ends a series of processes related to the correction function update. In addition, although the structure which performs the update of a correction function on the basis of the deviation width of a coordinate value was demonstrated here, this structure may be deform | transformed into the structure which performs the update of a correction function on the basis of the comparison result of a user characteristic. Alternatively, the correction function may be updated in a predetermined cycle.

  Here, the correction function update process in step S115 will be described in more detail with reference to FIG. As shown in FIG. 11, the function update unit 15 first detects a user operation pattern (S121). Next, the function updating unit 15 detects the size of the user's finger (S122). Next, the function update unit 15 transmits the detection processing results in steps S121 and S122 to the information processing system 100 (S123). Next, the function updating unit 15 receives a new correction function from the information processing system 100 (S124). Next, the function update unit 15 updates the correction function stored in the function storage unit 14 (S125), and ends the series of processes.

  Here, as an example, information on the user's operation pattern and the size of the user's finger is transmitted to the information processing system 100 and stored in the function storage unit 14 with a new correction function generated or selected according to the information. The configuration for updating the corrected function has been described. However, the type of user feature used for updating the correction function can be changed as appropriate. Further, the configuration can be changed so that the actually detected coordinate group is transmitted to the information processing system 100 instead of the user feature information.

  The operation of the information processing apparatus according to the present embodiment has been described above.

[2-3: Configuration of Information Processing System 100]
Next, the configuration of the information processing system 100 according to the present embodiment will be described with reference to FIGS. 12-16 is explanatory drawing for demonstrating the structure of the information processing system 100 which concerns on this embodiment.

  As illustrated in FIG. 12, the information processing system 100 includes a coordinate storage unit 101, a teacher data generation unit 102, a correction function generation unit 103, a feature amount extraction unit 104, a correction function classification unit 105, and a correction function provision. Unit 106 and user feedback reception unit 107. The coordinate storage unit 101 stores a group of coordinates detected in advance using the touch sensor 11.

First, the teacher data generation unit 102 acquires a coordinate group stored in the coordinate storage unit 101. Next, the teacher data generation unit 102 uses the acquired coordinate group as student data to generate teacher data. For example, as illustrated in FIG. 13, the teacher data generation unit 102 acquires coordinates X (t ₁ ) to X (t ₅ ) stored in the coordinate storage unit 101 as student data. Next, the teacher data generation unit 102 corrects the acquired coordinates X (t ₁ ) to X (t ₅ ) to generate the teacher data coordinates X ′ (t ₃ ). The type of correction used here corresponds to the type of correction by the correction function. For example, correction by moving average, correction by weighted average, correction by nonlinear fitting, correction by Butterworth filter, and the like can be used. In addition, considering the fact that the amount of noise varies from place to place due to the influence of the electrode structure and the like, the strength of the filter used for generating teacher data may be changed from place to place. By changing the strength of the filter for each location, teacher data for each location can be obtained. Therefore, the correction function generated using the teacher data is suitable for each location.

The teacher data generation unit 102 sequentially generates teacher data by the method shown in FIG. 13 while moving the range of student data used for generating teacher data. The teacher data generated by the teacher data generation unit 102 is input to the correction function generation unit 103. The student data used for generating the teacher data is input to the correction function generation unit 103 and the feature amount extraction unit 104. When teacher data and student data are input, the correction function generation unit 103 generates a correction function by machine learning using the input teacher data and student data. For example, as illustrated in FIG. 14, the correction function generation unit 103 generates a correction function F that maps the coordinates X (t ₁ ) to X (t ₅ ) of the student data to the coordinates X ′ (t ₅ ) of the teacher data. To do.

For example, consider the case where the function form of the correction function F is a linear function as shown in the following equation (1). In this case, the correction function generation unit 103 calculates a set of coefficients A _λ (λ = 1 to 5) that minimizes the square error Δ ² expressed by the following equation (2). The correction function generation unit 103 uses a coefficient A _λ (λ = 1 to ₅ ) for combinations of various student data coordinates X (t ₁ ) to X (t ₅ ) and teacher data coordinates X ′ (t ₅ ). ) Is calculated. Then, the correction function generation unit 103 generates the correction function F by substituting the calculated coefficient A _λ into the following equation (1). The correction function F generated by the correction function generation unit 103 is input to the correction function classification unit 105.

Although the case where the function form of the correction function F is a linear function has been described here, the same applies to the case where the function form of the correction function F is a nonlinear function. Further, although the correction function F for inputting the coordinates of five student data and outputting the coordinates of one teacher data has been described, the number of coordinates of the student data may be other than five. Further, as a method for calculating the coefficient A _λ, the method for minimizing the square error Δ ² has been described, but a method for minimizing the error Δ may be used. Here, although the method for calculating the coefficient regardless of the class described later has been described, the coefficient may be calculated for each class. A method for calculating the coefficient for each class will be described later. However, for convenience of explanation, it is assumed here that the correction function F is the linear function shown in the above equation (1), and the coefficient A _λ is obtained by a method that minimizes the square error Δ ^2. Proceed.

The feature amount extraction unit 104 acquires the coordinate group used as the student data from the coordinate storage unit 101, and classifies the acquired coordinate group into a predetermined class. First, the feature quantity extraction unit 104 calculates the speed of the operating tool at each coordinate using the acquired coordinate group. For example, the feature amount extraction unit 104 calculates a speed combination v = (v ₁ ,..., V ₅ ) related to the coordinates X (t ₁ ) to X (t ₅ ) of the student data. The feature quantity extraction unit 104, the coordinates _X (t 1) of the student data to X while moving the _{(t 5),} various student data coordinates _{X (t 1) ~X (t} 5) the rate combination with respect to v is calculated. The speed combination v calculated by the feature amount extraction unit 104 is input to the correction function classification unit 105.

Furthermore, the feature quantity extraction unit 104 extracts feature quantities other than the speed related to the coordinates X (t ₁ ) to X (t ₅ ) of the student data. For example, as shown in FIG. 15, the feature amount extraction unit 104 crosses the boundary of the grid whether the coordinates X (t ₁ ) to X (t ₅ ) of the student data indicate an operation in the grid of the touch sensor 11. It is determined whether the operation is indicated. Then, the feature amount extraction unit 104 inputs the determination result to the correction function classification unit 105 as a feature amount. When user feature information corresponding to each coordinate can be acquired, the feature amount extraction unit 104 corrects the user feature information corresponding to the coordinates X (t ₁ ) to X (t ₅ ) of the student data as the feature amount. Input to the function classification unit 105. For example, information (see FIG. 16) indicating whether the coordinate corresponds to the operation with the flat of the finger or the coordinate corresponding to the operation with the tip of the finger is input to the correction function classification unit 105 as a feature amount.

  When a feature amount is input, the correction function classification unit 105 classifies the correction function using the input feature amount. First, the correction function classification unit 105 classifies a combination of teacher data and student data (hereinafter referred to as learning data) using the velocity combination v. For example, the correction function classification unit 105 classifies learning data having a similar configuration of the speed combination v into the same class, and extracts representative learning data for each class. For example, the correction function classifying unit 105 obtains the centroid of the speed combination v corresponding to the learning data for each class, and extracts the learning data corresponding to the speed combination v closest to the centroid. Then, the correction function classification unit 105 selects a correction function corresponding to the representative learning data extracted for each class.

  Note that the correction function classification unit 105 may classify the correction function more finely using other feature amounts such as user features. In this case, the correction function classification unit 105 classifies the learning data using the speed combination v, and then classifies the learning data of each class into subclasses based on other feature amounts. Then, the correction function classification unit 105 extracts representative learning data for each subclass. For example, the correction function classifying unit 105 obtains the centroid of the speed combination v corresponding to the learning data for each subclass, and extracts the learning data corresponding to the speed combination v closest to the centroid. Then, the correction function classification unit 105 selects a correction function corresponding to the representative learning data extracted for each subclass.

  The correction functions classified by the correction function classification unit 105 in this way are stored in the function storage unit 14 of the information processing apparatus described above. By classifying as described above, the coordinate correction unit 13 of the information processing apparatus described above can use different correction functions according to the characteristics of the coordinate group detected by the touch sensor 11. For example, the coordinate correction unit 13 calculates a speed combination v ′ related to the coordinate group detected by the touch sensor 11 and selects a class corresponding to the speed combination v closest to the calculated speed combination v ′. Then, the coordinate correction unit 13 acquires a correction function corresponding to the selected class from the function storage unit 14, and calculates correction coordinates using the acquired correction function. When the correction function is classified into subclasses, the coordinate correction unit 13 is configured to calculate correction coordinates using a correction function of a subclass corresponding to another feature amount such as a detected user feature. It may be.

  When user feedback is transmitted to the information processing system 100, this user feedback is received by the user feedback receiving unit 107. For example, when a coordinate group is transmitted as user feedback, the user feedback receiving unit 107 inputs the received coordinate group to the teacher data generation unit 102, the correction function generation unit 103, and the feature amount extraction unit 104. When a coordinate group is input, the teacher data generation unit 102 generates teacher data by using the input coordinate group as student data. Further, the correction function generation unit 103 generates a correction function using the teacher data generated by the teacher data generation unit 102 and the coordinate group (student data). Furthermore, the feature amount extraction unit 104 extracts a feature amount from the input coordinate group. Then, the correction function classifying unit 105 classifies the correction function using the feature amount of the coordinate group, and inputs it to the correction function providing unit 106. The correction function providing unit 106 provides a correction function to the information processing apparatus that has transmitted the user feedback.

  In addition, information indicating a feature amount such as a user feature may be transmitted as user feedback from an information processing apparatus that does not hold correction functions classified into subclasses based on the feature amount such as a user feature. In this case, the correction function providing unit 106 selects a subclass corresponding to a feature quantity such as a user feature transmitted as user feedback, and provides a correction function belonging to the selected subclass to the information processing apparatus.

  The configuration of the information processing system 100 according to the present embodiment has been described above. In the above description, the configuration for classifying the correction function after generating the correction function has been described. However, the correction function is generated for learning data representing each class after classifying the learning data. It is also possible to deform. According to this modification, the amount of calculation required for generating the correction function can be reduced.

(How to calculate the coefficient for each class)
In the above method, the coefficient of the correction function is calculated using student data and teacher data, and the correction function is classified into classes. However, after classifying the coordinate groups used as student data, a correction function corresponding to each class may be generated using the coordinate groups classified for each class. For example, if the t-th student data is a vector vx _t = (x _t−4 , x _t−3 , x _t−2 , x _t−1 , x _t ) and the teacher data is x _t ′, the teacher data x _t ′ is expressed as x _t ′ = g (x _t−2 , x _t−1 , x _t , x _{t + 1} , x _{t + 2} ) using the correction function g.

At this time, if the student data vx _t is classified into the class c, the correction function g ^c, as in equation (3) below, is expressed by the coefficients A _j ^c corresponding to the class c. The coefficient A _j ^c corresponding to this class c is determined so as to minimize the square error Δ ² expressed by the following equation (4), for example. By this method, it is possible to calculate the correction function g ^c for each class. The method for classifying student data is as described above. When the correction function is generated by the method described here, the function of the correction function classification unit 105 in the information processing system 100 can be omitted. Further, the correction function generation unit 103 generates a correction function corresponding to each class using the coordinate group classified for each class by the feature amount extraction unit 104.

[2-4: Operation of information processing system 100]
Next, the operation of the information processing system 100 according to the present embodiment will be described with reference to FIGS. 17 to 19 are explanatory diagrams for explaining the operation of the information processing system 100 according to the present embodiment.

(Operation when not updated)
First, an operation when generating a correction function using a coordinate group stored in the coordinate storage unit 101 will be described.

  As shown in FIG. 17, first, the teacher data generation unit 102 acquires a coordinate group prepared in advance as student data from the coordinate storage unit 101 (S201). Next, the teacher data generation unit 102 generates teacher data using the student data acquired in step S201 (S202). Next, the correction function generation unit 103 generates a correction function by machine learning using the teacher data generated in step S202 and the student data acquired in step S201 (S203). Next, the correction function classification unit 105 classifies the correction function based on the feature amount extracted from the student data by the feature amount extraction unit 104 (S204), and selects a correction function that represents each class (S205). . Next, the correction functions classified by the correction function classification unit 105 are provided to the input device 10 (S206), and a series of processing ends.

(Operation # 1 when updating)
Next, an operation of the information processing system 100 in a situation where a coordinate group is transmitted from the input device 10 as user feedback will be described.

  As shown in FIG. 18, first, the user feedback receiving unit 107 receives a coordinate group as user feedback (S211). Next, the teacher data generation unit 102 sets the received coordinate group as student data (S212). Next, the teacher data generation unit 102 generates teacher data using the student data set in step S212 (S213). Next, the correction function generation unit 103 generates a correction function by machine learning using the teacher data generated in step S213 and the student data set in step S212 (S214). Next, the correction function classification unit 105 classifies the correction function based on the feature amount extracted from the student data by the feature amount extraction unit 104 (S215), and selects a correction function that represents each class (S216). . Next, the correction function classified by the correction function classification unit 105 is provided to the input device 10 (S217), and a series of processing ends.

(Operation # 2 when updating)
Next, an operation of the information processing system 100 in a situation where a user operation pattern is transmitted as user feedback will be described.

  As shown in FIG. 19, first, the user feedback receiving unit 107 receives a user operation pattern (S221). Next, the correction function providing unit 106 searches for a correction function corresponding to the user operation pattern received in step S221 (S222). Next, the correction function providing unit 106 determines whether there is a correction function corresponding to the user's operation pattern (S223). If there is a correction function corresponding to the user's operation pattern, the information processing system 100 advances the process to step S230. On the other hand, when there is no correction function corresponding to the user's operation pattern, the information processing system 100 advances the process to step S224.

  When the process proceeds to step S224, the user feedback reception unit 107 requests the input device 10 for a coordinate group corresponding to the user's operation pattern, and receives the coordinate group transmitted in response to the request (S224). ). Next, the teacher data generation unit 102 sets the received coordinate group as student data (S225). Next, the teacher data generation unit 102 generates teacher data using the student data set in step S224 (S226). Next, the correction function generation unit 103 generates a correction function by machine learning using the teacher data generated in step S226 and the student data set in step S225 (S227). Next, the correction function classification unit 105 classifies the correction function based on the feature amount extracted from the student data by the feature amount extraction unit 104 (S228), and selects a correction function that represents each class (S229). .

  When the process proceeds to step S230, the correction function classified by the correction function classification unit 105 is provided to the input device 10 (S230), and the series of processes ends. With such a configuration, even when a correction function corresponding to the user's operation pattern is not generated, it is possible to provide the input device 10 with a correction function that matches the operation pattern.

  The operation of the information processing system 100 according to the present embodiment has been described above.

[2-5: Summary of system configuration]
Up to now, the configuration of an information processing apparatus equipped with a function for updating a correction function has been described. However, when correction functions corresponding to a sufficient number of classes are prepared, the function of updating the correction function can be omitted. In the above description, the correction function is generated by the information processing system 100. However, a correction function generation function may be mounted on the information processing apparatus. Thus, the configuration of the information processing apparatus and the information processing system 100 can be changed as appropriate. Accordingly, a system configuration capable of realizing the technology according to the present embodiment without distinguishing between the information processing apparatus and the information processing system 100 will be briefly summarized.

(2-5-1: Configuration example 1)
First, refer to FIG. FIG. 20 is an explanatory diagram showing an example (configuration example 1) of a system configuration capable of realizing the technology according to the present embodiment. As shown in FIG. 20, the system according to Configuration Example 1 includes a touch sensor 51, a coordinate storage unit 52, a class classification unit 53, a coefficient storage unit 54, a coordinate correction unit 55, a switch 56, and teacher data. A generation unit 57, a teacher data storage unit 58, a student data storage unit 59, a coefficient calculation unit 60, and a switch 61 are included.

  For example, a touch sensor 51, a coordinate storage unit 52, a class classification unit 53, a coefficient storage unit 54, and a coordinate correction unit 55 are mounted on the information processing apparatus, and a teacher data generation unit 57, a teacher data storage unit 58, and a student data storage unit 59. The coefficient calculation unit 60 is mounted on the information processing system 100. Note that the switches 56 and 61 are schematically described in order to explain the operation of the system. Of course, components corresponding to the switches 56 and 61 may be actually provided in the system.

  First, the operation of the system when the switches 56 and 61 are off will be described.

  When an input operation is performed, coordinates are acquired by the touch sensor 51. The coordinates acquired by the touch sensor 51 are stored in the coordinate storage unit 52. The coordinates stored in the coordinate storage unit 52 are read out by the class classification unit 53. The class classification unit 53 detects the feature amount of the coordinate group read from the coordinate storage unit 52, and determines a class corresponding to the coordinate group based on the detected feature amount. For example, the class classification unit 53 calculates a speed at each coordinate read from the coordinate storage unit 52, and determines a class corresponding to the coordinate group read from the coordinate storage unit 52 based on the calculated combination of speeds.

  The class information determined by the class classification unit 53 is input to the coefficient storage unit 54. When class information is input, the coefficient storage unit 54 selects a coefficient corresponding to the input class from the coefficients of the correction function prepared for each class in advance, and the selected coefficient is used as the coordinate correction unit 55. To enter. When the coefficient is input, the coordinate correction unit 55 inputs the coordinate group read from the coordinate storage unit 52 to the correction function expressed using the input coefficient. Then, the coordinate correction unit 55 outputs the output value of the correction function as correction coordinates.

  The operation of the system when the switches 56 and 61 are off has been described above. When the coefficient of the correction function is prepared in advance for each class, the correction coordinates can be easily obtained by the above operation. For example, when a class corresponding to a feature close to the feature of the operation actually performed exists and a coefficient corresponding to the class is prepared, a sufficiently high correction accuracy can be obtained by the above configuration.

  Next, the operation of the system when the switches 56 and 61 are switched on will be described. That is, an operation when calculating a coefficient by machine learning will be described.

  First, the switch 56 is turned on, and the coordinate group classified by the class classification unit 53 is input to the teacher data generation unit 57 and the student data storage unit 59 as learning data. When this coordinate group is input, the teacher data generation unit 57 uses the input coordinate group as student data, and generates teacher data by the method shown in FIG. The teacher data generated by the teacher data generation unit 57 is stored in the teacher data storage unit 58. On the other hand, the coordinate group input to the student data storage unit 59 is held by the student data storage unit 59 as student data.

When teacher data is stored in the teacher data storage unit 58 and student data is stored in the student data storage unit 59, the coefficient calculation unit 60 calculates the coefficient of the correction function using these teacher data and student data. The teacher data stored in the teacher data 58 and the student data stored in the student data storage unit 59 are classified by class. Therefore, the coefficient for each class is calculated by the coefficient calculation unit 60 by using the teacher data and student data classified for each class. For example, the coefficient calculation unit 60 defines the correction function as in the above equation (3), and uses the teacher data x _t ^c and the student data vx _t corresponding to the class c to define in the above equation (4). The coefficient A _λ ^c is calculated so that the square error Δ _c ² is minimized. Next, the switch 61 is turned on, and the coefficient calculated by the coefficient calculation unit 60 is stored in the coefficient storage unit 54.

  The operation of the system when the switches 56 and 61 are turned on has been described above. For example, even when there is no class suitable for the operation pattern actually performed, the coefficient corresponding to the new class can be generated by the operation described here.

  Heretofore, an example (configuration example 1) of a system configuration capable of realizing the technology according to the present embodiment has been described.

(2-5-2: Configuration example 2)
Next, an example (configuration example 2) of a system configuration capable of realizing the technology according to the present embodiment will be described with reference to FIG. The difference from Configuration Example 1 shown in FIG. 20 is whether or not it is assumed that the coefficient of the correction function is updated when the accuracy of the correction function is low. As shown in FIG. 21, the system according to Configuration Example 2 includes a touch sensor 71, a coordinate storage unit 72, a class classification unit 73, a coefficient storage unit 74, a coordinate correction unit 75, and a teacher data generation unit 76. A teacher data storage unit 77, a student data storage unit 78, a divergence degree calculation unit 79, a switch 80, and a coefficient calculation unit 81.

  Hereinafter, the operation of the system according to Configuration Example 2 will be described.

  When an input operation is performed, coordinates are acquired by the touch sensor 71. The coordinates acquired by the touch sensor 71 are stored in the coordinate storage unit 72. The coordinates stored in the coordinate storage unit 72 are read out by the class classification unit 73. The class classification unit 73 detects the feature amount of the coordinate group read from the coordinate storage unit 72, and determines a class corresponding to the coordinate group based on the detected feature amount. For example, the class classification unit 73 calculates the speed at each coordinate read from the coordinate storage unit 72, and determines the class corresponding to the coordinate group read from the coordinate storage unit 72 based on the calculated combination of speeds.

  Information on the class determined by the class classification unit 73 is input to the coefficient storage unit 74. When the class information is input, the coefficient storage unit 74 selects a coefficient corresponding to the input class from the coefficients of the correction function prepared for each class in advance, and the selected coefficient is the coordinate correction unit 75. To enter. When the coefficient is input, the coordinate correction unit 75 inputs the coordinate group read from the coordinate storage unit 72 to the correction function expressed using the input coefficient. Then, the coordinate correction unit 75 outputs the output value of the correction function as correction coordinates.

  The coordinate group classified by the class classification unit 73 is input to the teacher data generation unit 76 and the student data storage unit 78 as learning data. When this coordinate group is input, the teacher data generation unit 76 uses the input coordinate group as student data, and generates teacher data by the method shown in FIG. The teacher data generated by the teacher data generation unit 76 is stored in the teacher data storage unit 77. On the other hand, the coordinate group input to the student data storage unit 78 is held by the student data storage unit 78 as student data.

  When teacher data is stored in the teacher data storage unit 77 and student data is stored in the student data storage unit 78, the divergence degree calculation unit 79 calculates the coordinates of the teacher data and the coordinates of the student data corresponding to the coordinates. The degree of divergence between them is calculated. For example, the divergence degree calculation unit 79 extracts teacher data and student data for each class, and calculates the average divergence degree of each class. Then, the divergence degree calculating unit 79 determines whether or not the calculated divergence degree exceeds a predetermined threshold value. When the divergence degree exceeds a predetermined threshold, the divergence degree calculation unit 79 turns on the switch 80 and inputs the teacher data and student data of the class corresponding to the divergence degree to the coefficient calculation unit 81.

When the teacher data and the student data are input, the coefficient calculation unit 81 calculates the coefficient of the correction function using the input teacher data and the student data. For example, when the class corresponding to the divergence degree exceeding a predetermined threshold is class c, the coefficient calculation unit 81 defines the correction function as in the above equation (3), and the input teacher data x _t ^c and The coefficient A _λ ^c is calculated using the student data vx _{t so} that the square error Δ _c ² defined by the above equation (4) is minimized. The coefficient calculated by the coefficient calculation unit 81 is stored in the coefficient storage unit 74.

  Heretofore, an example (configuration example 2) of a system configuration capable of realizing the technology according to the present embodiment has been described. With such a configuration, the coefficient of the correction function is updated when the accuracy of the correction function is low, such as when there is no class suitable for the actual operation pattern.

<3: Hardware configuration>
The function of each component included in the information processing apparatus and the information processing system 100 can be realized by using, for example, a part or all of the hardware configuration illustrated in FIG. That is, the function of each component is realized by controlling the hardware shown in FIG. 22 using a computer program. The form of the hardware is arbitrary, and includes, for example, a personal computer, a mobile phone, a portable information terminal such as a PHS, a PDA, a game machine, or various information appliances. However, the above PHS is an abbreviation of Personal Handy-phone System. The PDA is an abbreviation for Personal Digital Assistant.

  As shown in FIG. 22, this hardware mainly includes a CPU 902, a ROM 904, a RAM 906, a host bus 908, and a bridge 910. Further, this hardware includes an external bus 912, an interface 914, an input unit 916, an output unit 918, a storage unit 920, a drive 922, a connection port 924, and a communication unit 926. However, the CPU is an abbreviation for Central Processing Unit. The ROM is an abbreviation for Read Only Memory. The RAM is an abbreviation for Random Access Memory.

  The CPU 902 functions as, for example, an arithmetic processing unit or a control unit, and controls the overall operation of each component or a part thereof based on various programs recorded in the ROM 904, the RAM 906, the storage unit 920, or the removable recording medium 928. . The ROM 904 is a means for storing a program read by the CPU 902, data used for calculation, and the like. In the RAM 906, for example, a program read by the CPU 902, various parameters that change as appropriate when the program is executed, and the like are temporarily or permanently stored.

  These components are connected to each other via, for example, a host bus 908 capable of high-speed data transmission. On the other hand, the host bus 908 is connected to an external bus 912 having a relatively low data transmission speed via a bridge 910, for example. As the input unit 916, for example, a mouse, a keyboard, a touch panel, a button, a switch, a lever, or the like is used. Further, as the input unit 916, a remote controller (hereinafter referred to as a remote controller) capable of transmitting a control signal using infrared rays or other radio waves may be used.

  As the output unit 918, for example, a display device such as a CRT, LCD, PDP, or ELD, an audio output device such as a speaker or a headphone, a printer, a mobile phone, or a facsimile, etc. Or it is an apparatus which can notify audibly. However, the above CRT is an abbreviation for Cathode Ray Tube. The LCD is an abbreviation for Liquid Crystal Display. The PDP is an abbreviation for Plasma Display Panel. Furthermore, the ELD is an abbreviation for Electro-Luminescence Display.

  The storage unit 920 is a device for storing various data. As the storage unit 920, for example, a magnetic storage device such as a hard disk drive (HDD), a semiconductor storage device, an optical storage device, a magneto-optical storage device, or the like is used. However, the HDD is an abbreviation for Hard Disk Drive.

  The drive 922 is a device that reads information recorded on a removable recording medium 928 such as a magnetic disk, an optical disk, a magneto-optical disk, or a semiconductor memory, or writes information to the removable recording medium 928. The removable recording medium 928 is, for example, a DVD medium, a Blu-ray medium, an HD DVD medium, or various semiconductor storage media. Of course, the removable recording medium 928 may be, for example, an IC card on which a non-contact type IC chip is mounted, an electronic device, or the like. However, the above IC is an abbreviation for Integrated Circuit.

  The connection port 924 is a port for connecting an external connection device 930 such as a USB port, an IEEE 1394 port, a SCSI, an RS-232C port, or an optical audio terminal. The external connection device 930 is, for example, a printer, a portable music player, a digital camera, a digital video camera, or an IC recorder. However, the above USB is an abbreviation for Universal Serial Bus. The SCSI is an abbreviation for Small Computer System Interface.

  The communication unit 926 is a communication device for connecting to the network 932. For example, a wired or wireless LAN, Bluetooth (registered trademark), or a WUSB communication card, an optical communication router, an ADSL router, or various types It is a modem for communication. The network 932 connected to the communication unit 926 is configured by a wired or wireless network, such as the Internet, home LAN, infrared communication, visible light communication, broadcast, or satellite communication. However, the above LAN is an abbreviation for Local Area Network. The WUSB is an abbreviation for Wireless USB. The above ADSL is an abbreviation for Asymmetric Digital Subscriber Line.

<4: Summary>
Finally, the technical idea of this embodiment will be briefly summarized. The technical idea described below can be applied to various information processing apparatuses such as a PC, a mobile phone, a portable game machine, a portable information terminal, an information home appliance, and a car navigation system.

The functional configuration of the information processing apparatus described above can be expressed as follows. The information processing apparatus includes a noise removal unit and a function generation unit described below. Among the plurality of coordinates acquired by the touch sensor, the noise removing unit includes the nth acquired coordinate _Xn and the (n−p) th (p = 1 to P) acquired coordinate X. _The coordinate X from which the noise corresponding to the coordinate _Xn is removed using _(n−p) and the coordinate X _(n−q) acquired in the (n + q) th (q = 1 to _{Q). n} 'is calculated. In addition, the function generation unit described above is a function that outputs the coordinate X _n ′ when the (n−m) th (m = 1 to M) acquired coordinate X _(nm) is input. Fn is generated by machine learning.

When the function F _n is used, it is possible to remove the noise at the coordinate X _n using the coordinate group already held at the timing when the coordinate X _n is acquired. For example, if it is attempted to remove the noise at the _nth coordinate Y _n acquired in the same manner as the noise removal unit without using the function F _n , the (n + Q) th coordinate Y _{(n + Q)} is It is necessary to wait until it is acquired. That is, a delay occurs for the time from when the coordinate Y _n is acquired until the coordinate Y _{(n + Q)} is acquired. However, when the function F _n is used, noise can be removed at the timing when the coordinate Y _n is acquired, and thus the delay does not occur. That is, if the function _Fn is used, it is possible to remove coordinate noise in real time. As a result, the coordinate detection accuracy can be improved without deteriorating the response.

(Remarks)
The information processing system 100 is an example of a coordinate correction function generation device. The teacher data generation unit 102 is an example of a noise removal unit. The correction function generation unit 103 is an example of a function generation unit. The feature amount extraction unit 104 is an example of a speed calculation unit and a region determination unit. The correction function classifying unit 105 is an example of a function classifying unit. The user feedback reception unit 107 is an example of a coordinate information acquisition unit. The function update unit 15 is an example of an operation pattern detection unit, a function selection unit, an operation tool size detection unit, and a deviation degree detection unit.

  The preferred embodiments according to the present technology have been described above with reference to the accompanying drawings, but it is needless to say that the present technology is not limited to the configuration examples disclosed herein. It will be apparent to those skilled in the art that various changes and modifications can be made within the scope of the claims, and these are naturally within the technical scope of the present technology. Understood.

  For example, among the system configurations described so far, matters relating to system design, such as which components are installed in the information processing apparatus and which components are installed in the information processing system 100, are determined according to the embodiment. It is desirable to decide appropriately. For example, as shown in FIG. 6, a configuration for updating the correction function may be mounted on the information processing apparatus (or the input device 10), or a configuration for updating the correction function may be omitted. Of the components shown in FIG. 20, the touch sensor 51, the coordinate storage unit 52, the class classification unit 53, the coefficient storage unit 54, and the coordinate correction unit 55 may be mounted on the information processing device (or input device). It is. Of course, it goes without saying that such modifications are also included in the technical scope of the present embodiment.

DESCRIPTION OF SYMBOLS 10 Input device 11 Touch sensor 12 Coordinate storage part 13 Coordinate correction part 14 Function storage part 15 Function update part 20 Information processing part 30 Display part 51, 71 Touch sensor 52, 72 Coordinate storage part 53, 73 Class classification part 54, 74 Coefficient Storage unit 55, 75 Coordinate correction unit 56, 61, 80 Switch 57, 76 Teacher data generation unit 58, 77 Teacher data storage unit 59, 78 Student data storage unit 60, 81 Coefficient calculation unit 79 Deviation degree calculation unit 100 Information processing system DESCRIPTION OF SYMBOLS 101 Coordinate storage part 102 Teacher data generation part 103 Correction function generation part 104 Feature-value extraction part 105 Correction function classification | category part 106 Correction function provision part 107 User feedback receiving part

Claims (18)

Of the plurality of coordinates acquired by the touch sensor, the nth acquired coordinate _Xn , the (np) th (p = 1 to P) acquired coordinate X _(np), and And (n + q) th (q = 1 to Q) acquired coordinates X _(n−q), and noise for calculating coordinates X _n ′ from which noise corresponding to the coordinates X _n has been removed A removal section;
A function for generating, by machine learning, a function F _{n from} which the coordinates X _n ′ are output when the coordinates _(n−m) acquired in the (n−m) th (m = 1 to M) are input. A generator,
Comprising
Coordinate correction function generator. The function generation unit expresses the function F _n by a linear combination of the coordinates X _(n−m) (m = 1 to M), and the coordinates X _n ′ and the coordinates X _(n−m) (m = 1 to M) to calculate the coupling coefficient by the method of least squares using the actually measured values.
The coordinate correction function generator according to claim 1. The noise removing unit averages the coordinates _Xn , the coordinates X _(n−p) (p = 1 to P), and the coordinates X _{(n + q)} (q = _{1 to Q)} to obtain the coordinates X _n ′ is calculated,
The coordinate correction function generator according to claim 2. The noise removing unit is configured to multiply the coordinate X _n and a weight D _p (D _p <1) that decreases as p increases to the coordinate X _(n−p) , and a value as q increases. The coordinate X _n ′ is calculated by averaging the weight D _q (D _q <1) that is smaller than the value obtained by multiplying the coordinate X _{(n + q)} by
The coordinate correction function generator according to claim 2. The noise removing unit inputs the coordinates _Xn , the coordinates X _(np) (p = 1 to P), and the coordinates X _{(n + q)} (q = 1 to Q) to a Butterworth filter. To calculate the coordinates X _n ′,
The coordinate correction function generator according to claim 2. The noise removing unit is a regression line indicating the relationship between the coordinate _Xn , the coordinate X _(n−p) (p = 1 to P), and the coordinate X _{(n + q)} (q = _{1 to Q)} or A regression curve is calculated, and a coordinate on the regression line or regression curve corresponding to the coordinate X _n is defined as the coordinate X _n ′.
The coordinate correction function generator according to claim 2. A speed calculating unit for calculating a moving speed at the coordinate _Xn ;
A function classifying unit that classifies the function F _n based on the moving speed calculated by the speed calculating unit, and generates a function representing each class;
Further comprising
The coordinate correction function generator according to claim 3. A speed calculating unit for calculating a moving speed at the coordinate _Xn ;
An area determination unit that divides the grid of the touch sensor into a plurality of areas and determines an area including the coordinate _Xn ;
A function classifying unit that classifies the function F _n based on the moving speed calculated by the speed calculating unit and the region determined by the region determining unit, and generates a function representing each class;
Further comprising
The coordinate correction function generator according to claim 3. A user uses the touch sensor from an input device that includes the touch sensor and a coordinate correction unit that inputs the coordinates acquired by the touch sensor to the function generated by the function generation unit and corrects the coordinates. A coordinate information acquisition unit for acquiring information related to the coordinates input in
The noise removal unit calculates coordinates from which noise has been removed using a coordinate group indicated by the information acquired by the coordinate information acquisition unit,
The function generation unit generates a function using the coordinate group indicated by the information acquired by the coordinate information acquisition unit, and the coordinates calculated by the noise removal unit,
The coordinate correction function generator according to claim 1. The coordinate information acquisition unit acquires information indicating an operation pattern of the user as information on the coordinates,
The noise removing unit calculates coordinates from which noise has been removed using a coordinate group corresponding to the operation pattern,
The function generation unit generates a function using a coordinate group corresponding to the operation pattern and the coordinates calculated by the noise removal unit.
The coordinate correction function generator according to claim 9. A touch sensor for detecting touched coordinates;
Of the plurality of coordinates acquired in advance by the touch sensor, the nth acquired coordinate _Xn and the (n−p) th (p = 1 to P) acquired coordinate X _{(n−p) )} and, based on the (n + q) th (q = 1 to q) coordinates _X obtained in _(n-q) and the coordinates _{X n} 'corresponding to the coordinates _{X n} in which noise is removed using, A function F _n that outputs the coordinate X _n ′ when the (n−m) th (m = 1 to M) acquired coordinate X ₍ nm) generated by machine learning is input. In contrast, a coordinate correction unit that inputs coordinates detected by the touch sensor and corrects the coordinates;
Comprising
Input device. An operation pattern detection unit that detects an operation pattern of a user using the touch sensor;
A function selection unit that selects a function corresponding to the operation pattern detected by the operation pattern detection unit from among the functions prepared for each predetermined operation pattern;
Further comprising
The coordinate correction unit corrects the coordinates by inputting the coordinates detected by the touch sensor for the function selected by the function selection unit.
The input device according to claim 11. An operating tool size detection unit that detects the size of the operating tool that operates the touch sensor;
A function selection unit that selects a function corresponding to the size of the operation object detected by the operation object size detection unit from among the functions prepared for each operation object size;
Further comprising
The coordinate correction unit corrects the coordinates by inputting the coordinates detected by the touch sensor for the function selected by the function selection unit.
The input device according to claim 11. A divergence degree detection unit that detects a divergence degree between the coordinates detected by the touch sensor and the coordinates corrected by the coordinate correction unit;
The coordinate correction unit changes the function when the divergence degree detected by the divergence degree detection unit is larger than a predetermined threshold.
The input device according to claim 11. Of the plurality of coordinates acquired by the touch sensor, the nth acquired coordinate _Xn , the (np) th (p = 1 to P) acquired coordinate X _(np), and And (n + q) th (q = 1 to Q) acquired coordinates X _(n−q), and noise for calculating coordinates X _n ′ from which noise corresponding to the coordinates X _n has been removed A removal step;
A function for generating, by machine learning, a function F _{n from} which the coordinates X _n ′ are output when the coordinates _(n−m) acquired in the (n−m) th (m = 1 to M) are input. Generation step;
including,
Coordinate correction function generation method. Of the plurality of coordinates acquired by the touch sensor, the nth acquired coordinate _Xn , the (np) th (p = 1 to P) acquired coordinate X _(np), and And (n + q) th (q = 1 to Q) acquired coordinates X _(n−q), and noise for calculating coordinates X _n ′ from which noise corresponding to the coordinates X _n has been removed A removal function;
A function for generating, by machine learning, a function F _{n from} which the coordinates X _n ′ are output when the (n−m) th (m = 1 to M) acquired coordinates X _(nm) are input. Generation function,
A program to make a computer realize. A coordinate detection step in which the touched coordinates are detected by the touch sensor;
Of the plurality of coordinates acquired in advance by the touch sensor, the nth acquired coordinate _Xn and the (n−p) th (p = 1 to P) acquired coordinate X _{(n−p) )} and, based on the (n + q) th (q = 1 to q) coordinates _X obtained in _(n-q) and the coordinates _{X n} 'corresponding to the coordinates _{X n} in which noise is removed using, A function F _n that outputs the coordinate X _n ′ when the (n−m) th (m = 1 to M) acquired coordinate X ₍ nm) generated by machine learning is input. On the other hand, a coordinate correction step for inputting the coordinates detected by the touch sensor and correcting the coordinates,
including,
Coordinate correction method. A coordinate detection function for detecting a touched coordinate by a touch sensor;
Of the plurality of coordinates acquired in advance by the touch sensor, the nth acquired coordinate _Xn and the (n−p) th (p = 1 to P) acquired coordinate X _{(n−p) )} and, based on the (n + q) th (q = 1 to q) coordinates _X obtained in _(n-q) and the coordinates _{X n} 'corresponding to the coordinates _{X n} in which noise is removed using, A function F _n that outputs the coordinate X _n ′ when the (n−m) th (m = 1 to M) acquired coordinate X ₍ nm) generated by machine learning is input. On the other hand, a coordinate correction function for inputting the coordinates detected by the touch sensor and correcting the coordinates,
A program to make a computer realize.

Images