CN101607138B - Action recognition method based on finite automata model - Google Patents

Action recognition method based on finite automata model Download PDF

Info

Publication number
CN101607138B
CN101607138B CN2008100435135A CN200810043513A CN101607138B CN 101607138 B CN101607138 B CN 101607138B CN 2008100435135 A CN2008100435135 A CN 2008100435135A CN 200810043513 A CN200810043513 A CN 200810043513A CN 101607138 B CN101607138 B CN 101607138B
Authority
CN
China
Prior art keywords
operation
dynamic
sub
data
action
Prior art date
Application number
CN2008100435135A
Other languages
Chinese (zh)
Other versions
CN101607138A (en
Inventor
张静盛
Original Assignee
盛乐信息技术(上海)有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 盛乐信息技术(上海)有限公司 filed Critical 盛乐信息技术(上海)有限公司
Priority to CN2008100435135A priority Critical patent/CN101607138B/en
Publication of CN101607138A publication Critical patent/CN101607138A/en
Application granted granted Critical
Publication of CN101607138B publication Critical patent/CN101607138B/en

Links

Abstract

本发明公开了一种基于有限自动机模型的动作识别方法,包括以下步骤:定义子动作;把动感操作划分为若干子动作,用动作表达式描述其子动作序列;把加速度传感器采集的动感数据划分为若干连续的数据子集,把各数据子集判断为相应的子动作,得到动感数据的子动作序列;引入有限自动机模型,根据动感数据的子动作及动感操作的动作表达式,进行动作识别。 The present invention discloses a motion recognition method based on finite automata model, comprising the steps of: defining a sub operation; the dynamic operation is divided into several sub-operation, which is described by the expression subaction operation sequence; dynamic data collected by the acceleration sensor divided into a number of successive subsets of data, each of the data subsets corresponding sub operation is determined to give the sub-dynamic action-sequence data; the finite automata model, and the dynamic operation in accordance with the operation of the sub-action expression dynamic data, for Action recognition. 本发明的动作识别方法,能实时地处理输入动作数据,提升动作识别效率。 Action recognition method of the present invention, the input operation can process data in real time, to enhance the efficiency of operation of identification.

Description

基于有限自动机模型的动作识别方法 Action recognition method based on finite automaton model

技术领域 FIELD

[0001] 本发明涉及动作感应技术,特别涉及一种基于有限自动机模型的动作识别方法。 [0001] The present invention relates to a motion sensing technology, particularly to the action recognition method based on finite state machine model. 背景技术 Background technique

[0002] 一般情况下,根据动作幅度的大小可以把动作感应技术分为两种:微小动作的感应和大幅度的动作感应,常见的微小动作的感应有如飞行模拟器等,如果手柄微小的变化,在游戏中都能够感受到;常见的大幅度动作感应有如任天堂2006年发布的电视游戏机Wii If small changes in handle fine motion sensing and motion-sensing substantial common induction minute operation like a flight simulator: [0002] In general, according to the size of the range of motion of the motion sensing technology can be divided into two in the game can feel; like a common substantial motion-sensing Nintendo video game released in 2006 Wii

等,大幅度动作如上/下/左/右等,通常手臂会有动作。 And so on, a significant action as up / down / left / right, etc., are usually the arm will be action.

[0003] 动作感应技术的应用需要结合相应的硬件来实现,以游戏手柄为例,对于第一种微小动作感应技术,手柄中需要有能够测定旋转速度的陀螺仪,在软件端只需要把陀螺仪测定的旋转角速度进行简单的处理即可。 [0003] Application of motion sensing technology requires a combination of appropriate hardware implementation to handle the game as an example, for the first minute motion sensing technology, the handle need be measured rotational speed gyroscope, the software only need to end the gyro rotational angular velocity meter measurement process can be simple.

[0004] 对于大幅度动作感应技术,相应的硬件至少需要测定XYZ三轴线性加速度,在此基础还可以增加测定旋转角速度的陀螺仪。 [0004] The motion sensing technology for dramatically, the corresponding hardware is necessary to measure at least XYZ-axis linear acceleration, on the basis of the rotational angular velocity may also increase the measured gyroscope.

[0005] 大幅度动作感应技术的动感分析算法的是以动感手柄的3轴或5轴加速度传感器感应值为输入,通过软件算法最终分析出用户动感操作。 [0005] substantial dynamic motion sensing technology is based on dynamic analysis algorithm handle 3 or 5-axis acceleration sensor for sensing an input value, in the final analysis software algorithm dynamic operation of the user.

[0006]目前大幅动感分析算法还处于起步阶段,一般只能是识别一些的简单动作,例如上、下、左、右、前、后等。 [0006] At present a substantial dynamic analysis algorithm still in its infancy, usually only recognizes a few simple actions, such as up, down, left, right, front, back and so on. 对于一些复杂动作的识别都需要花费大量时间和精力去识别,而且每添加一个动作都是如此。 For some complex actions identified need to spend a lot of time and effort to identify, and each add an action is true. 而且算法中存在大量的浮点运算,严重影响效率,延迟响应时间。 And there are a lot of floating-point algorithms, seriously affecting the efficiency, response time delay.

发明内容 SUMMARY

[0007] 本发明要解决的技术问题是提供一种基于有限自动机模型的动作识别方法,采用该动作感应方法,能实时地处理输入动作数据,提升动作识别效率。 [0007] The present invention is to solve the technical problem of providing a method for motion recognition model based on finite state machine, using the motion sensing method, capable of handling real-time data input operation, the motion recognition improve efficiency.

[0008] 为解决上述技术问题,本发明的基于有限自动机模型的动作识别方法,包括以下步骤: [0008] In order to solve the above problems, the action recognition method based on finite state machine model of the present invention, comprising the steps of:

[0009] (I)定义子动作; [0009] (I) defined in sub-operation;

[0010] (2)把动感操作划分为若干子动作,用动作表达式描述其子动作序列; [0010] (2) The dynamic operation is divided into several sub-actions, an operation expression used to describe its operation promoter sequence;

[0011] (3)把加速度传感器采集的动感数据划分为若干连续的数据子集,把各数据子集判断为相应的子动作,得到动感数据的子动作序列; [0011] (3) the collection of dynamic data of the acceleration sensor is divided into a plurality of contiguous data subsets, the subset of data is determined for each respective sub-action, action-sequence to give the sub-dynamic data;

[0012] (4)引入有限自动机模型,根据动感数据的子动作及动感操作的动作表达式,进行动作识别。 [0012] (4) the finite automata model, based on the operation expression subaction dynamic data and dynamic operations, perform action recognition.

[0013] 步骤(3)得到的动感数据的子动作序列可以先进行误差屏蔽后再进行步骤(4)。 Subaction sequence obtained [0013] Step (3) dynamic data error masking can be performed after the first step (4).

[0014] 可以根据动感数据的子动作的空间位移量、平均加速度进行误差屏蔽。 [0014] according to the amount of spatial displacement subaction dynamic data, the average acceleration error masking.

[0015] 本发明的基于有限自动机模型的动作识别方法,引入有限自动机(简称FA)模型,可以识别复杂的动作。 [0015] The method of the present invention, the motion recognition model based on finite state machine, the finite automaton (referred to as FA) model, complex actions can be identified. 对于复杂动作,把它分割为若干连续的子动作,以子动作序列表示这个实际的动作。 For complex operation, it is divided into a number of consecutive sub-operation, the sub-sequence representing the actual operation of the operation. 把这个序列逐一输入到FA,FA根据自己的内部逻辑经过一定的状态转换,输出动作判定结果。 This sequence by one input to the FA, FA after a certain state transitions according to their internal logic, the determination result output operation. 能实时地处理输入动作数据,提升动作识别效率。 Can be processed in real time data input operation, the motion recognition improve efficiency.

附图说明 BRIEF DESCRIPTION

[0016] 下面结合附图及具体实施方式对本发明作进一步详细说明。 [0016] Hereinafter, the present invention is described in further detail in conjunction with accompanying drawings and specific embodiments.

[0017] 图1是本发明的基于有限自动机模型的动作识别方法一实施方式流程图; [0017] FIG. 1 is an operation based on recognition of the finite state machine model of the present invention, a flow diagram embodiment;

[0018] 图2是一动感操作“顺时针圆”示意图; [0018] FIG. 2 is a schematic view of a dynamic operation "circle clockwise";

[0019] 图3是一子动作定义方式示例; [0019] FIG. 3 is a subaction defined by way of example;

[0020] 图4是一动感操作划分子动作示例; [0020] FIG. 4 is an exemplary dynamic operation subaction divided;

[0021] 图5是对“顺时针圆”的二种子动作划分方法示例。 [0021] FIG. 5 is an example of the "circle clockwise" in the operation of two seeds division method.

具体实施方式 Detailed ways

[0022] 本发明的基于有限自动机模型的大幅动作感应方法一实施方式如图1所示,包括以下步骤: [0022] The operation of the present invention is substantially based on sensing method of a finite automaton model of the embodiment 1, comprising the steps of:

[0023] (I)定义子动作; [0023] (I) defined in sub-operation;

[0024] (2)把动感操作划分为若干子动作,在配置文件中用数学表达式描述其子动作序列; [0024] (2) The dynamic operation is divided into several sub-operation in the configuration file using mathematical expressions which describe the operation of the sub-sequence;

[0025] (3)把加速度传感器采集的动感手柄的动感数据划分为若干连续的数据子集,把各数据子集判断为相应的子动作,得到动感数据的子动作序列; [0025] (3) collecting the dynamic acceleration sensor handle dynamic data into a plurality of contiguous data subsets, each subset of data to the respective sub-action is determined, the operation of the sub-sequence to obtain the dynamic data;

[0026] (4)把根据动感数据判断得到的子动作序列根据子动作的空间位移量、平均加速度进行误差屏蔽,以消除动感手柄静态感应和微小抖动所带来的误差; [0026] (4) According to the operation sequence of the sub determination of dynamic data obtained based on the amount of spatial displacement subaction, the average acceleration error masking in order to eliminate errors dynamic and static induction handle jitter caused by minute;

[0027] (5)引入有限自动机模型,根据动感数据的子动作及在配置文件中用数学表达式描述的动感操作子动作序列,进行动作识别。 [0027] (5) the finite automata model, according subaction and dynamic data in dynamic operation sub-profile operation using the mathematical expression described sequence, for recognition operation.

[0028] 本发明的基于有限自动机模型的动作识别方法,引入有限自动机(简称FA)模型,可以识别复杂的动作。 [0028] The method of the present invention, the motion recognition model based on finite state machine, the finite automaton (referred to as FA) model, complex actions can be identified. 对于复杂动作,把它分割为若干连续的子动作,以子动作序列表示这个实际的动作。 For complex operation, it is divided into a number of consecutive sub-operation, the sub-sequence representing the actual operation of the operation. 把这个序列逐一输入到FA,FA根据自己的内部逻辑经过一定的状态转换,输出动作判定结果。 This sequence by one input to the FA, FA after a certain state transitions according to their internal logic, the determination result output operation. 能实时地处理输入动作数据,提升动作识别效率。 Can be processed in real time data input operation, the motion recognition improve efficiency.

[0029] 无论是3轴还是5轴的加速度传感器,其动作感应是以空间3轴加速度的形式出现的,这就意味着当动感手柄在空间作匀速运动时,其没有感应值,此时与静止时无异。 [0029] Whether or 3-axis acceleration sensor 5-axis, which motion sensor is in the form of three-axis acceleration space appears, this means that when the handle is in a dynamic spatial uniform motion, which is not sensed value, and at this time when still no different. 本发明的技术方案中,会把无感应值或者只有微小感应的动作忽略。 Aspect of the present invention will no or only a slight induction value induced motion ignored. 本发明适用于3维空间的大幅动作感应,但为了便于精确地描述本发明的技术,下面以2维动作平面中的“顺时针圆”为例,对本发明的技术方案进行描述。 The present invention is applicable to 3-dimensional motion sensing significantly, but for convenience to accurately describe the techniques of this invention, the following operation of a two-dimensional plane "clockwise round" for example, the technical solution of the present invention will be described.

[0030] 1.定义子动作 [0030] 1. Definitions subaction

[0031] 在平面中可以定义8个子动作,分别是“一7丨丨\”。 [0031] 8 may be defined in sub-operation plane, they are "a Shushu 7 \." 但这并不是唯一的定义,也可以定义为4个、12个甚至更多,主要是由应用程序对动感操作精确性的不同需求而定。 But this is not the only definition that can also be defined as 4, 12 or even more, mainly by the application of dynamic operational needs of accuracy may be different. 子动作的多种定义方式如图3所示。 Subaction various definitional manner as shown in FIG.

[0032] 对于本发明的基于有限自动机模型的动作识别方法来说,定义的子动作越多,能够识别的动作就越细致、越精确。 [0032] The operation of recognition based on finite state machine model of the present invention, the more defined sub-operation, more detailed operation can be recognized more accurately. 但在实际实现过程中,例如游戏动感手柄,由于动感手柄硬件输出参数的误差性,以及PC的实际运算能力等各方面因素的限制,这个精确度就会受到影响。 However, in the actual implementation process, such as games handle dynamic, due to various error factors of the output parameters of the dynamic handle hardware, PC and the like of the actual computing capacity, the accuracy will be affected. 因而不能无限制地过多定义子动作。 And therefore can not be unlimited. Define too much action. 这里假设定义为8个。 It is assumed here defined as eight. [0033] 2.把动感操作划分为若干个子动作 [0033] 2. The dynamic operation of the operation into several sub

[0034] 动作感应技术的一个根本性问题是“如何识别动感操作”。 [0034] A fundamental problem is the motion sensing technology "how to identify dynamic operation." 在本发明中,基于第I点中所定义的子动作,把整个动感操作划分为若干连续的子动作。 In the present invention, the first operation based on the sub I as defined in point, the whole dynamic operating sub-divided into several continuous operation. 最后再对这个子动作序列进行匹配。 Finally, this sub-sequence of actions to match.

[0035] 划分动感操作的一个通用的方法是基于时间来划分,假设动感操作如图4所示,图4中的曲线表示动感操作的轨迹,其中的2条虚线把这个操作划分为3段,即划分为3个子动作。 A general method [0035] The dynamic operation is divided based on time division, assuming dynamic operation trace as shown in FIG 4 a graph showing the dynamic operation of FIG 4, wherein the broken line 2 this operation is divided into three segments, That is divided into three sub operation. 根据第I点中所定义的子动作来看,可以把图3中的3个阶段依次判定为“一7—”。 The first subaction I as defined in point of view, can be put in order in three stages in FIG. 3 is determined as "a 7-." 对这个动感操作进行描述的动作表达式预先定义存放在配置文件中,当用户动感手柄做一动作后,加速度传感器采集的动感手柄的动感数据能够与配置文件中预先定义的动作表达式“一/—”匹配,就被认为是目标动作,应用程序就读取配置文件中动作表达式的描述,然后动感地建立FA,进而进行动作识别,执行相应的逻辑功能。 This dynamic action expression of predefined operations described in the configuration file is stored, when the user handles do a dynamic operation, dynamic acceleration sensor data acquisition can handle dynamic profile with a predefined operation expression "a / - "match, target operation is considered, the application reads the operation expression described in the configuration file, then the FA build dynamic, and further operates to recognize and execute the appropriate logic function. 本发明的方法并不一定需要动感操作完全的一致,这个过程中有一定的容错性。 The method of the present invention is not necessarily exactly coincide dynamic operation, this process has certain fault tolerance.

[0036] 例如对于图2中所描述的顺时针圆就可以进行如图5所示划分: [0036] For example, a circle clockwise in FIG. 2 described can be divided as shown in FIG 5:

[0037] 图5中有两种划分: [0037] FIG. 5 is divided in two:

[0038] (I)左边:划分为4个子动作,可以依次识别为“ ,W。 [0038] (I) left: divided into four sub-operation may be sequentially identified as a ", W.

[0039] (2)右边:划分为4个子动作,可以依次识别为“一丨一t ”。 [0039] (2) Right: divided into four sub-operation may be sequentially identified as "a Shu a T."

[0040] 这两种划分形式,都可以作为识别的标准。 [0040] The two division form, it can be used as the identification criterion.

[0041] 但从动感操作“顺时针圆”本身来说,不仅仅是这两种。 [0041] However, the dynamic operation is "circle clockwise" itself, not just two. 对于上面的每一种划分其实可以细化为4种类型: For each of the above can be divided in fact subdivided into four types:

[0042] (I)①一②一③一④ [0042] (I) ① - ② ③ a a ④

[0043] (2)②一③一④一① [0043] (2) ② ③ a a a ④ ①

[0044] (3)③—④—①—② [0044] (3) ③-④-①-②

[0045] (4)④—①—②—③ [0045] (4) ④-①-②-③

[0046] 那么图5中所代表的就是8种子动作序列。 [0046] So as represented in FIG. 5 is an operation sequence 8 seeds. 用户动感手柄的动作只要能够与其中之一种匹配,那就可以识别为“顺时针圆”动作。 Dynamic operation of the user as long as the handle is matched with one of which, it can be recognized as "circle clockwise" action. 这些内容在实际开发过程中都可用配置文件形式呈现,而在程序内部需要实现识别这些配置动作表达式的引擎。 The content in the actual development process are presented in the form of available profiles, and in the need to implement internal procedures to identify these engine configuration actions expressions.

[0047] 本发明的这种动作识别方法除了能够很方便地识别复杂动作外,还有另外一个优势。 [0047] This action recognition method of the present invention, in addition able to easily identify complex operation, there is another advantage. 假设现在需要识别“η次顺时针圆”这个操作,那么只需要简单修改配置动作表达式即可,例如对于“2次顺时针圆”可以表不为“I — 2 — 3 — 4 — 1 — 2 — 3 — 4”。 Suppose now need to identify "[eta] Ci circle clockwise" This operation, then only need to modify the configuration simple operation expression can, for example, "circle twice clockwise" table may not "I - 2 - 3 - 4 - 1 - 2--3--4. "

[0048] 3.把动感数据划分为若干连续的数据子集 [0048] 3. The dynamic data into several successive subset of the data

[0049] 前面两个步骤,其实是为分析动感操作搭建分析框架。 [0049] The preceding two steps, in fact, is to build dynamic analysis framework for analyzing operation. 从这一步开始,就可以进行具体的数据分析与动作识别。 From this point, we can carry out specific data analysis and action recognition.

[0050] 前面也已经提到对于动作的识别,首先要把动感数据根据时间进行分段。 [0050] The foregoing has referred to the identified operation, the dynamic data should first segment of time. 在分段判定动感数据的子动作之后,再对动感数据的连续的子动作序列进行识别。 After the segment is determined subaction dynamic data, then the continuous operation of sub-sequences identified dynamic data.

[0051] 动感手柄的加速度传感器对于动作感应的输出是在X,y, ζ三轴方向上的加速度,记为(ax,ay, az),因而首先需要把这些数据转化为对应的空间运动轨迹的点集。 The acceleration sensor [0051] The dynamic operation of the handle to the output induced in the X, y, ζ accelerations in three axes, referred to as (ax, ay, az), and therefore first need to be converted to the corresponding data space trajectory the set of points.

[0052] 假设采样为100次/S,子动作分段时间为50ms,如果用于完成“顺时针圆”这个动作需要250ms,那么可以划分为5个子动作的数据子集。 [0052] assumed that the sampling of 100 times / S, subaction segment time is 50ms, if a complete "circle clockwise" This action requires 250ms, then the operation can be divided into five sub-subset of data. 然后对每个子动作的数据集合进行线性拟合(因为在上面第二点中的子动作都定义为线性的,在运算速度上具有很大优势)。 And operation data of each sub-set of the linear fit (subaction second point as above defined are linear, the computing speed has great advantages). [0053] 4.屏蔽静态感应和微小抖动所带来的误差 [0053] 4. A static induction shielding fine jitter and error caused by

[0054] 在动感手柄进行动感操作过程中难免有些误操作,例如用户握住手柄保持不动,但实际上很难确保完全静止,或多或少加速度传感器总会有些感应值。 [0054] The dynamic operation of the process is inevitable that some erroneous operation in the dynamic handle, such as a user holds the handle remains stationary, but in practice difficult to ensure a complete standstill, there is always more or less the acceleration sensor sensing value. 因而对于这一类的误差的处理是非常有必要的。 Thus for this type of error handling it is necessary.

[0055] 在一些应用程序中对于动作感应的需求是大幅度类型的动作,那么用户如有少量小型的动作也会被列入屏蔽的范围。 [0055] In some applications the need for substantial motion sensing is the type of action, the user action also small if a small amount was included in the scope of the shield.

[0056] 另外还有一种情况,当一个动感操作结束时,从空间运动轨迹上来说,几乎为静止。 [0056] There is another case, when the end of a dynamic operation, from the spatial trajectory is almost stationary. 但从力学中动能定理的角度来看,动作结束时会有一个感应值比较大的反向冲力,在动感数据中表现的就是一系列感应值比较大的值。 But from the perspective mechanics kinetic energy theorem of view, there is a relatively large inductance reverse the momentum of the end of the action, which is a series of relatively large inductance value in the performance of dynamic data. 这也是需要去屏蔽的。 This is also the need to shield.

[0057] 在本发明中,可以根据子动作进行误差屏蔽,大致有两种方式: [0057] In the present invention, error masking can be performed in accordance with the sub-action, there are basically two ways:

[0058] (I)子动作的空间位移量 [0058] The amount of displacement space (I) subaction

[0059] (2)子动作的平均加速度 [0059] The average acceleration (2) subaction

[0060] 5.引入有限自动机(FA)模型识别动感操作 [0060] The introduction of finite automata (FA) dynamic model identification operation

[0061] 有限自动机是计算机理论中,能够合理有效地处理复杂状态转换逻辑的一种模型,其形式定义如下5个部分:状态集合、字母表、转移函数、起始状态和终止状态。 [0061] The finite state machine is a computer model theory, it is possible to reasonably and effectively handle complex logic state transitions, which form part 5 is defined as follows: set of states, an alphabet, the transfer function, the initial and final states.

[0062] 在本发明中引入有限自动机模型,与一般的动作感应技术相比有以下几个优势: [0062] In the finite automaton model of the present invention, the general motion sensing technology has several advantages as compared to:

[0063] (I)实时地处理输入。 [0063] (I) processes the input in real time.

[0064] 通常的动作感应技术,在处理感应动感数据时需要等到动感操作完成时,才能对数据进行处理。 [0064] The conventional motion sensing technology, when the dynamic sensor data processing need to wait until the completion of the dynamic operation, to process the data. 而在本发明中,极大地提升动作识别技术中的实时数据处理能力,每一个子动作的产生都可以作为FA的输入,实时地改变FA的内部状态。 In the present invention, the motion recognition technology greatly enhance the real-time data processing capability, each sub-generating operation can be used as an input of the FA, to change the internal state of the real FA.

[0065] (2)提升动作识别效率。 [0065] (2) improve the efficiency of operation of identification.

[0066] FA识别动作所需要的时间取决于子动作的数量。 Time [0066] FA identify an action required depends on the number of sub-operation. 例如在完成一个动感操作需要250ms,每个子动作时间为50ms,那么每隔50ms就会产生一个子动作信号。 For example, in a dynamic operation requires complete 250ms, the operation time of each sub-50ms, it will produce a signal subaction every 50ms. 对于FA来说,每隔50ms会接收一个输入的子动作信号,从而驱使一系列状态的变化。 For FA, the subaction every 50ms receives a signal inputted to drive state change of series.

[0067] (3)有效地组织各类动感操作。 [0067] (3) efficient organization of various dynamic operation.

[0068] 对于每个动感操作,由于其本身存在多种情况,如第2点中所提到的“顺时针圆”这个操作。 [0068] For each dynamic operation, since there are many cases in itself, such as the second point referred to "circle clockwise" is operating. 一般情况下,识别时会逐个进行匹配。 In general, one by one match is identified. 但在本发明中引入FA模型,就不需要了。 However, the present invention is introduced in the FA model, it is not required. 由于“顺时针圆”这个操作划分为4个子动作,那么对于这个动作的识别,就是由子动作数量决定,固定为4,与这个操作有几种情况无关。 As the "circle clockwise" This operation is divided into four sub-action, then the action for the identification, is determined by the number subaction fixed to 4, there is nothing to do with this operation several situations.

[0069] (4)简化开发流程。 [0069] (4) simplify the development process.

[0070] 在具体实现过程中,首先采用动作表达形式描述动感操作,即描述其子动作序列,然后在内部实现构建FA的引擎。 [0070] In a specific implementation, the first use of an operation expression described in the form of dynamic operation, i.e., sub-actions which sequence is described, and constructed to achieve the FA within the engine. 应用程序首先读取配置文件中对于动感操作的描述的动作表达形式,然后动感的建立FA,接下去就可以进行动作识别。 Application first reads the operation expressed in the form of the profile of the description of dynamic operation, then the FA dynamic establishment, operation next can be identified.

[0071] 当需要增加对新的动感操作的识别时,只需要在配置文件中添加该动作的表达式描述即可,而不需要修改任何模块。 [0071] when it is necessary to increase the recognition of new dynamic operation, only you need to add the operation expression described in the configuration file can, without any need to modify the module.

Claims (3)

1.一种基于有限自动机模型的动作识别方法,其特征在于,包括以下步骤: (1)定义子动作; (2)把动感操作划分为若干子动作,用动作表达式描述其子动作序列; (3)把加速度传感器采集的动感数据划分为若干连续的数据子集,把各数据子集判断为相应的子动作,得到动感数据的子动作序列; (4)引入有限自动机模型,根据动感数据的子动作及动感操作的动作表达式,进行动作识别。 A motion recognition method based on finite state machine model, characterized by comprising the steps of: (1) defining a sub operation; (2) The dynamic operation is divided into several sub-actions, an operation expression used to describe its sequence subaction ; (3) the dynamic acceleration data acquired by the sensor is divided into a plurality of contiguous data subsets, the respective data subset is determined respective sub-action, to give the sub-sequences of actions to dynamic data; and (4) the finite automata model, according to subaction dynamic action expression data and dynamic operation, operates identification.
2.根据权利要求1所述的基于有限自动机模型的动作识别方法,其特征在于,把步骤(3)得到的动感数据的子动作序列进行误差屏蔽后再进行步骤(4)。 The action recognition model based on the finite state machine of claim 1, wherein the step (3) to give the sub-dynamic operation sequence data after the error masking step (4).
3.根据权利要求2所述的基于有限自动机模型的动作识别方法,其特征在于,根据动感数据的子动作的空间位移量、平均加速度进行误差屏蔽。 The action recognition model based on the finite state machine of claim 2, wherein, based on the amount of spatial displacement subaction dynamic data, the average acceleration error masking.
CN2008100435135A 2008-06-17 2008-06-17 Action recognition method based on finite automata model CN101607138B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN2008100435135A CN101607138B (en) 2008-06-17 2008-06-17 Action recognition method based on finite automata model

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN2008100435135A CN101607138B (en) 2008-06-17 2008-06-17 Action recognition method based on finite automata model

Publications (2)

Publication Number Publication Date
CN101607138A CN101607138A (en) 2009-12-23
CN101607138B true CN101607138B (en) 2013-11-20

Family

ID=41481146

Family Applications (1)

Application Number Title Priority Date Filing Date
CN2008100435135A CN101607138B (en) 2008-06-17 2008-06-17 Action recognition method based on finite automata model

Country Status (1)

Country Link
CN (1) CN101607138B (en)

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103218062A (en) * 2013-04-24 2013-07-24 伍斌 Man-machine interaction method and equipment based on acceleration sensor and motion recognition
CN109189218A (en) * 2018-08-20 2019-01-11 广州市三川田文化科技股份有限公司 A kind of method, apparatus of gesture identification, equipment and computer readable storage medium

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1153576A (en) 1995-05-10 1997-07-02 任天堂株式会社 Operating device with analog joystick
CN1474424A (en) 1994-05-09 2004-02-11 索尼计算机娱乐公司 Controller unit for electronic device
CN1770064A (en) 2005-07-15 2006-05-10 中国海洋大学 Interactive input device for computer
CN1996208A (en) 2007-01-15 2007-07-11 解晖 Man-machine command input device and mapping method for motion information in the same

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1474424A (en) 1994-05-09 2004-02-11 索尼计算机娱乐公司 Controller unit for electronic device
CN1153576A (en) 1995-05-10 1997-07-02 任天堂株式会社 Operating device with analog joystick
CN1770064A (en) 2005-07-15 2006-05-10 中国海洋大学 Interactive input device for computer
CN1996208A (en) 2007-01-15 2007-07-11 解晖 Man-machine command input device and mapping method for motion information in the same

Also Published As

Publication number Publication date
CN101607138A (en) 2009-12-23

Similar Documents

Publication Publication Date Title
JP6064280B2 (en) System and method for recognizing a gesture
Cohen et al. Convergence properties of the gravitational algorithm in asynchronous robot systems
KR920006283B1 (en) Digital signal processing method
US4228498A (en) Multibus processor for increasing execution speed using a pipeline effect
US20060229861A1 (en) Multi-core model simulator
CN102016616B (en) Combination logic randomness
Smith et al. The design and analysis of parallel algorithms
CN103430148A (en) State grouping for element utilization
Gu et al. Human gesture recognition through a kinect sensor
US20130120282A1 (en) System and Method for Evaluating Gesture Usability
CN102985897A (en) Efficient gesture processing
Killingbeck Microcomputer Quantum Mechanics: 0
JPH09282080A (en) Information input method/device
US20100283739A1 (en) Contact grouping and gesture recognition for surface computing
KR101873619B1 (en) Boolean logic in a state machine lattice
KR101185144B1 (en) Method and apparatus for estimating 2-dimension trajectory of a gesture
Mueller et al. Fpga: What's in it for a database?
JP3924256B2 (en) Compiler apparatus, compiler program, recording medium, compiling, runtime information generating device, and runtime information generating program
JP2007280129A (en) Similar shape search device, similar shape search method, and program
JP3971085B2 (en) Swept volume model
CN100401267C (en) Chip dynamic tracing method of microprocessor
JP4137434B2 (en) Simulation method and simulation method and recorded computer-readable recording medium a program
Grigorian et al. BRAINIAC: Bringing reliable accuracy into neurally-implemented approximate computing
CN105765528A (en) Application execution path tracing with configurable origin definition
US7110934B2 (en) Analysis of the performance of a portion of a data processing system

Legal Events

Date Code Title Description
C06 Publication
C10 Entry into substantive examination
C14 Grant of patent or utility model
TR01 Transfer of patent right