Optical motion mode sensing system and method

Abstract

The present invention relates to an optical motion mode sensing system and method. The system comprises an image sensor, an image processing module and a motion detection module. The method comprises: collection of a dynamic image, processing of the image, generation and sensing of a specific event cloud, and state conversion of a state machine. The optical motion mode sensing system and method disclosed in the present invention realizes sensing of a moving object by means of an image sensor sensing a light intensity change caused by object motion, and thus, compared with traditional recognition techniques, the requirement for a higher motion precision of human-machine interaction developing towards artificial intelligence can be met; moreover, compared with new recognition modes, the optical motion mode sensing system and method disclosed in the present invention has no requirement for extracting basic information and processing a background, thereby reducing the data volume; and a binary or low-precision digital signal can be generated on the basis of the sensing of the light intensity change, thereby greatly reducing the computing power, lowering the implementation threshold, and fitting many Internet of Things application programs.

Description

Optical motion pattern perception system and method Technical Field

[0001] The present invention belongs to the field of motion perception technology, and in particular relates to an optical motion pattern perception system and method. Background Art

[0002] Human-computer interaction (HCI) has been a hot topic in the development of artificial intelligence (AI) in recent years, and motion sensing or recognition technology is a crucial foundation for HCI. Currently, common motion sensing or recognition technologies can be divided into two categories: traditional recognition methods such as infrared thermal sensing and radio frequency identification (RFID), and newer recognition methods such as facial recognition that require feature extraction and deep learning algorithms.

[0003] Infrared thermal sensing relies on infrared radiation emitted by an object to detect obstacles within a predetermined area. Radio frequency identification (RFID) technology requires a radio frequency card that transmits radio frequency signals within a specific area for sensing, and it operates only within the range of the transmitted radio frequency signal. Both infrared sensing and RFID, as traditional recognition methods, can only detect or recognize objects when they are stationary or moving at low speeds. They cannot detect objects moving within space, and therefore cannot perceive moving objects. Because traditional recognition technologies lack the ability to generate images and accurately judge based on images, their application scenarios are limited, making them unsuitable for the higher motion accuracy required by the evolution of human-computer interaction towards artificial intelligence. Newer recognition methods, which combine hardware and software, capture and further process images to establish spatial motion correlations, improving recognition reliability and are currently gaining popularity. Taking facial recognition as an example, facial recognition requires pre-collection of basic information, such as facial features, and extraction of feature values ​​from key facial points. Deep learning software (such as widely used neural network learning algorithms) is then used to further process the facial images. Emerging recognition methods rely heavily on collected baseline data and require increasingly complex algorithms to train on this data, resulting in massive datasets and further increasing the complexity of these algorithms. In pursuit of reliability, these recognition methods require not only ever-increasing amounts of data but also increasingly complex algorithms and computational power. This undoubtedly raises the bar for implementation and increases the difficulty of implementation. For many IoT applications, which require solutions that require less data and are easier to implement, these new recognition methods are less versatile.

[0004] It can be seen from this that traditional recognition technologies such as infrared thermal sensing and wireless radio frequency identification cannot be applied to the needs of human-computer interaction developing towards artificial intelligence, which requires higher motion accuracy, because they cannot perceive moving targets. New recognition methods represented by face recognition focus on software and neglect hardware, resulting in increasing data volume, increasingly complex algorithms, higher computing power, and increasing difficulty in implementation, which in turn limits their application.

[0005] Summary of the Invention

[0006] In response to the above problems, the purpose of the present invention is to provide an optical motion pattern perception system and method, which is used to solve the problem that traditional recognition technologies such as infrared thermal sensing and wireless radio frequency identification cannot perceive moving targets and cannot be applied to the needs of human-computer interaction developing towards artificial intelligence, which requires higher motion accuracy. New recognition methods represented by face recognition focus on software and neglect hardware, resulting in increasing amounts of data, increasingly complex algorithms, higher computing power, and increasing difficulty in implementation, which in turn limits their application.

[0007] To achieve the above object, the present invention adopts the following technical solutions:

[0008] In a first aspect, the present invention discloses an optical motion pattern sensing system, comprising

[0009] At least one image sensor that captures images in real time, where changes in light intensity trigger pixel events within its pixels, and where all pixel events of the image sensor form an output that constitutes an event cloud;

[0010] At least one image processing module predefines a path on the image and predefines a number of specific areas on the path. A trigger pixel threshold is set for each specific area. When the number of pixels changed in a specific area exceeds the trigger pixel threshold, the specific area is triggered to form a specific event cloud.

[0011] at least one motion detection module, comprising at least one state machine, the state machine having N states, where N is a positive integer greater than 1, the state machine selecting one state from a set of N states as an initial state, the state machine being controlled by a digital signal generated by the image sensor in response to a specific event cloud, and generating a plurality of control signals based on the state of the state machine;

[0012] Among them, the image sensor collects images in real time and generates a dynamic event cloud; the event cloud moves on the image along the path predefined by the image processing module, and passes through these specific areas in sequence according to the time series. If the number of pixels changed in a specific area exceeds the trigger pixel threshold, the specific area is triggered to form a specific event cloud; the image sensor senses the specific event cloud according to the change in light intensity and generates a binary or low-precision digital signal, and transmits the digital signal to the motion detection module; until the timing relationship between the specific event clouds triggered by these specific areas meets the specific requirements of the event cloud in the specific area of ​​the path, the motion detection module generates multiple control signals according to the state that needs to change in the state machine to realize the state transition of the state machine.

[0013] Furthermore, a timing monitor is included, which can record the time of a specific event cloud in real time and is associated with the state machine of the motion inspection module, and can return the state machine to an initial state;

[0014] Once the state of the state machine encounters a deadlock phenomenon or a timeout phenomenon, the timing monitor returns the state machine to an initial state according to the control signal or timeout information generated by the motion inspection module.

[0015] Specifically, the image sensor is a motion image sensor, which includes a pixel matrix and a pixel control circuit, each pixel has a pixel structure including a photodetector and a storage unit, the pixel control circuit is capable of sequentially writing photodetection values ​​representing after exposure to the storage unit of the pixel, and the image sensor also includes a comparison circuit, which is used to store the sequential photodetection values ​​of at least part of the pixels, to compare the sequential photodetection values, to generate a comparison signal based on the sequential photodetection values ​​of the pixels, and to generate a comparison matrix reflecting image changes from the comparison signal.

[0016] Specifically, the image processing module and the motion detection module are integrated on the image sensor.

[0017] Furthermore, a controller is included, and the image sensor, the image processing module, the motion detection module and the timing monitor are independent of each other and are respectively connected to the controller.

[0018] Furthermore, it includes multiple image sensors, which are arranged in a manner to expand the field of view of motion perception as much as possible; the multiple image sensors are connected to the controller in sequence or separately; each image sensor is configured with an image processing module, and each image sensor predefines a path on the image captured by the corresponding image sensor to form multiple paths, and predefines several specific areas on each path, and the several specific areas of the multiple paths are associated with one or more state machines.

[0019] Furthermore, it includes multiple image processing modules, which are respectively connected to the image sensor or respectively connected to the controller; the multiple image processing modules predefine paths on the image captured by an image sensor to form multiple paths, and predefine several specific areas on each path, and the several specific areas of the multiple paths are associated with one or more state machines.

[0020] In a second aspect, the present invention discloses an optical motion pattern perception method, comprising:

[0021] Dynamic image acquisition: The image sensor acquires images in real time. Changes in light intensity trigger pixel events within its pixels. The output formed by all pixel events of the image sensor constitutes an event cloud.

[0022] Image processing: The image processing module predefines a path on the image and predefines a number of specific areas on the path;

[0023] Generation and perception of a specific event cloud: The event cloud moves along a path on the image and passes through specific areas in a time sequence. If the number of pixels changing in a specific area exceeds its trigger pixel threshold, the specific area is triggered to form a specific event cloud. The image sensor senses the specific event cloud based on changes in light intensity and generates a binary or low-precision digital signal, which is then transmitted to the motion detection module.

[0024] State transition of the state machine: until the timing relationship between the specific event clouds triggered by these specific areas meets the specific requirements of the event clouds in the specific areas of the path, the motion inspection module generates multiple control signals according to the state that needs to change in the state machine to realize the state transition of the state machine.

[0025] Furthermore, while the image sensor senses the specific event cloud based on the light intensity change and generates a binary or low-precision digital signal, and transmits the digital signal to the motion detection module, the image sensor also transmits the digital signal to a timing monitor, and the occurrence time of the specific event cloud is recorded in real time by the timing monitor;

[0026] Furthermore, once the state of the state machine encounters a deadlock phenomenon or a timeout phenomenon, the timing monitor returns the state machine to an initial state according to the control signal or timeout information generated by the motion inspection module.

[0027] Specifically, the predefined path is: a straight line or an arc in which the hand of the person being sensed swings from the left to the right or from the right to the left;

[0028] The predefined several specific areas are: a first specific area and a second specific area are predefined on the left and right sides of the path respectively;

[0029] The timing relationship between the specific event clouds triggered by these specific areas meets the specific requirements of the event cloud in the specific areas of the path, thereby realizing the state conversion of the state machine: the event cloud passes through the first specific area and the first specific area along the path. If the first specific area and the first specific area are triggered in sequence according to the chronological order, forming the first specific event cloud and the second specific event cloud respectively, then starting from the initial state, the first specific event cloud is generated first and then the second specific event cloud is generated, then the state machine generates one state; conversely, starting from the initial state, the second specific event cloud is generated first and then the first specific event cloud is generated, then the state machine generates another state.

[0030] Specifically, the predefined path is: the route that the perceived person's feet walk from outside the door to inside the door;

[0031] The predefined several specific areas are: according to the foot walking from the outside to the inside in chronological order, a third specific area, a fourth specific area and a fifth specific area are predefined in sequence from the outside to the inside on the path;

[0032] The timing relationship between the specific event clouds triggered by these specific areas meets the specific requirements of the event cloud in the specific areas of the path, thereby realizing the state conversion of the state machine: the event cloud passes through the third specific area, the fourth specific area and the fifth specific area along the path. If the third specific area, the fourth specific area and the fifth specific area are triggered in sequence according to the chronological order, forming the third specific event cloud, the fourth specific event cloud and the fifth specific event cloud respectively, then starting from the initial state, the third specific event cloud is generated first, then the fourth specific event cloud is generated, and finally the fifth specific event cloud is generated, and the state of the state machine is converted.

[0033] Specifically, the predefined path is: the route that the perceived person walks from outside the window through the window to inside the window;

[0034] The predefined specific areas are: after the perceived person passes through the window from outside to inside, and then moves away from the window in chronological order, a sixth specific area, a seventh specific area, and an eighth specific area are predefined in sequence on the path;

[0035] The timing relationship between the specific event clouds triggered by these specific areas meets the specific requirements of the event cloud in the specific areas of the path, thereby realizing the state conversion of the state machine: the event cloud passes through the sixth specific area, the seventh specific area and the eighth specific area along the path. If the sixth specific area, the seventh specific area and the eighth specific area are triggered in sequence according to the chronological order, forming the sixth specific event cloud, the seventh specific event cloud and the eighth specific event cloud respectively, then starting from the initial state, the sixth specific event cloud is generated first, then the seventh specific event cloud is generated, and finally the eighth specific event cloud is generated, and the state of the state machine is converted.

[0036] In a third aspect, the present invention further discloses the application of the above-mentioned optical motion pattern sensing method in waving switches, automatic doors or window intrusion alarms.

[0037] The present invention has the following advantages due to the adoption of the above technical solution:

[0038] (1) The present invention proposes an optical motion pattern perception system and method, wherein an image sensor captures images in real time, and changes in light intensity trigger pixel events within its pixels. The output formed by all pixel events is defined as an event cloud; an image processing module predefines a path on the image and predefines several specific areas on the path; the event cloud moves along the path on the image and passes through these specific areas in sequence according to a time sequence. If the number of pixels that change in a specific area exceeds its triggering pixel threshold, the specific area is triggered to form a specific event cloud; the image sensor senses the specific event cloud based on the change in light intensity and generates a binary or low-precision digital signal, which is transmitted to a motion detection module and a timing monitor. The occurrence time of the specific event cloud is recorded in real time by the timing monitor; until the timing relationship between the specific event clouds triggered by these specific areas meets the specific requirements of the event cloud in the specific area of ​​the path, the motion detection module generates multiple control signals based on the state of the state machine that needs to change, thereby realizing the state transition of the state machine. In the event that the state machine encounters a deadlock or timeout phenomenon, the timing monitor returns the state machine to the initial state based on the control signal or timeout information generated by the motion detection module. The optical motion pattern perception system and method disclosed in the present invention achieves the perception of moving targets by using an image sensor to sense changes in light intensity caused by the object's motion. Compared with traditional recognition technologies such as infrared thermal sensing and wireless radio frequency identification, this system and method can meet the needs of human-computer interaction towards artificial intelligence, which requires higher motion accuracy. Furthermore, unlike new recognition methods such as facial recognition, which focus on software and light on hardware, the optical motion pattern perception system and method disclosed in the present invention firstly eliminates the need to extract basic information and process background information, thus reducing the amount of data. Secondly, it can generate binary or low-precision digital signals based on the perception of light intensity changes, greatly reducing computing power and thus lowering the threshold for implementation, ultimately enabling a wide range of applications.

[0039] (2) The emerging recognition technology combining hardware and software has shown a development trend of emphasizing software and neglecting hardware, and the "wrong" cognition gradually generated by this development trend in the field of artificial intelligence, that is, the larger the data, the more complex the algorithm, and the more accurate the recognition result. Therefore, in order to obtain accurate recognition results, it is necessary to further increase the research on deep learning algorithms. However, this development idea actually increases the difficulty of implementing emerging recognition technologies, thereby limiting their promotion and application. The present invention discloses an optical motion pattern perception system and method. According to the light intensity change caused by the movement of an object, the light intensity change caused by the movement of the object is sensed by an image sensor, and the non-moving background is automatically removed, thereby realizing the perception of the moving target. There is no need to extract basic information and process the background, which reduces the amount of data; it can generate binary or low-precision digital signals based on the perception of light intensity changes, greatly reducing the demand for computing power and further reducing complexity. Therefore, compared with the emerging recognition method, the present invention provides a technical solution with a technical concept that is completely different from the existing sensing and recognition technology. Its motion recognition accuracy is high, and the technical effect reaches or even exceeds that of the existing emerging recognition technology, overcoming the problems in recent years caused by the development of emphasizing software and neglecting hardware. The technical bias formed by the "wrong" cognition that has gradually emerged in the field of artificial intelligence is due to the trend. That is, it is possible to obtain highly accurate motion recognition without extracting significant information and reducing the required computing power. Furthermore, as the Internet of Things applications have an increasing demand for smaller-capacity, easier-to-implement motion recognition and human-computer interaction solutions, the present invention may cause researchers and engineers in the fields of motion perception, human-computer interaction and even artificial intelligence to rethink the development trend of heavy software and light hardware presented by the emerging recognition technology combining hardware and software, causing relevant researchers to pay attention to hardware, and guiding more people to invest more research on the sensing of motion image sensors based on changes in light intensity caused by the movement of objects.

[0040] (3) The present invention discloses an optical motion pattern perception system and method, wherein the optical motion pattern perception system can be used as a motion-based human-computer interaction device and can be implemented by a simple motion image sensor and a simple hardware or software time state machine; since moving objects usually only occupy a small part of the scene, for example, waving hands, foot movements, etc., this motion can be perceived by an image sensor that is sensitive to motion, obtaining a sparse and binary image, greatly reducing the amount of data and increasing the useful information content of the data, which makes the optical motion pattern perception system and method disclosed in the present invention require less data and is easier to implement, further lowering the implementation threshold. The optical motion pattern perception system and method disclosed in the present invention have a wide range of application scenarios, such as gesture control, motion control, gait sensing, number of people, contactless door opening, etc. For many IOT (Internet of Things) applications that require smaller data volumes and cheaper solutions, the present invention can be universally applicable. Whether compared with traditional recognition technologies and emerging recognition technologies, or in terms of the technical effects of recognition and sensing, or even in terms of promotion and application, many aspects have confirmed that the present invention provides a fast, efficient and economical solution based on motion perception. Moreover, from the current perspective, whether in terms of perception devices or perception methods, the optical motion pattern perception system and method disclosed in the present invention can be considered to be highly optimized in human-computer interaction configuration.

[0041] (IV) The present invention discloses an optical motion pattern sensing system and method, which employs an image sensor having a novel sensor architecture, including a pixel matrix and a pixel control circuit. Each time an image is captured, a pixel matrix of the same size as the image is generated. The pixel control circuit encodes the image changes in a simple data format based on the local temporal light intensity changes of each pixel in two consecutive images, and generates a pixel-by-pixel output. For example, the pixel values ​​of pixels at the same position in two consecutive images are compared. If there is no change, a NULL value is output. If there is a change, a positive output is activated if the pixel value increases (increases in brightness), and a negative output is activated if the pixel value decreases (decreases in brightness). In order to display it more intuitively, it can be assumed that the color representing the output NULL value is white, the color representing the positive output is red, and the color representing the negative output is blue. Then the optical motion pattern perception system and method disclosed in the present invention, the image sensor of which displays to the user the image: the part where the light intensity changes presents a dynamic alternation of red and blue, and the part where the light intensity does not change becomes the background, which can be statically presented in colors such as white. The area that needs to display the change is reduced, further reducing the amount of data; at the same time, since static photos are not directly generated, and only dynamic images that display the local light intensity change part are relied upon, the target person cannot be effectively identified. Therefore, the present invention provides privacy protection for people who are regarded as motion perception objects. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] FIG1 is a working principle diagram of an optical motion pattern perception system disclosed in Example 1 of the present invention, including an image sensor and an image processing module;

[0043] FIG2 is a working principle diagram of the optical motion pattern perception system disclosed in Example 1 of the present invention, which includes an image sensor and two image processing modules;

[0044] FIG3 is a diagram illustrating an application example of a waving switch in the optical motion pattern sensing method disclosed in Example 2 of the present invention, wherein FIG3(I) is a schematic diagram of pre-defining a path and a specific area on an image, and FIG3(II) is a corresponding state machine diagram;

[0045] FIG4 is a diagram illustrating an application example of an automatic door in the optical motion pattern sensing method disclosed in Example 2 of the present invention, wherein FIG4(I) is a schematic diagram of pre-defining a path and a specific area on an image, and FIG4(II) is a corresponding state machine diagram;

[0046] FIG5 is a diagram showing an application example of the window intrusion alarm in the optical motion pattern perception method disclosed in Example 2 of the present invention, wherein FIG5(I) is a schematic diagram of predefined paths and specific areas on an image, and FIG5(II) is a corresponding state machine diagram.

[0047] Explanation of the accompanying drawings: a-image sensor, b-image processing module, c-motion detection module, d-timing monitor; 100-image, 101-path, 102-specific area; A1-first specific area, B1-first specific area, A2-third specific area, B2-fourth specific area, C2-fifth specific area, C3-sixth specific area, B3-seventh specific area, A3-eighth specific area; ZA1-first specific event cloud, ZB1-second specific event cloud, ZA2-third specific event cloud, ZB2-fourth specific event cloud, ZC2-fifth specific event cloud, ZC3-sixth specific event cloud, ZB3-seventh specific event cloud, ZA3-eighth specific event cloud. DETAILED DESCRIPTION

[0048] Exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present invention are shown in the accompanying drawings, it should be understood that the present invention can be implemented in various forms and should not be limited by the embodiments described herein. Rather, these embodiments are provided to enable a more thorough understanding of the present invention and to fully convey the scope of the present invention to those skilled in the art.

[0049] Emerging recognition methods that combine hardware and software, such as facial recognition, are striving for reliability by increasing data volumes and algorithms, requiring ever-higher computing power. This has led to a trend of prioritizing software over hardware. This excessive focus on algorithmic sophistication has fostered a misconception within the field of artificial intelligence: the larger the data volume and the more complex the algorithm, the more accurate the recognition results. This has even fostered a bias against simplifying significant information and reducing computing power to achieve accurate results. This bias has substantially limited the development of motion sensing technology.

[0050] Studies have shown that the movement of objects usually causes changes in light intensity, which can be perceived by optical sensors. Therefore, it is feasible to perceive moving targets through optical-based sensors.

[0051] Because of the "event" perception of light, the background can be considered "erased" (some noise may appear on the background due to noise in the sensor), and each moving object generates a cloud of "trigger" pixels in the event image. In a well-designed human-computer interaction system, the complexity of the motion pattern can be simplified to achieve simplicity and increase reliability. However, this "event" image (local light intensity changes are considered to be triggered by some event) must be further processed.

[0052] To simplify the extraction of significant information and reduce the required computing power, some smart sensing pixel arrays have been proposed, such as the "Improved CMOS Image Sensor." The Chinese invention patent application number 202210132353.1, "Improved CMOS Image Sensor," proposes some smart sensing pixel arrays. These smart sensing pixel arrays can provide an image in which each pixel value is sensitive to local temporal light intensity changes, generating binary or low-precision digital signals based on light intensity changes. These digital signals are usually two-valued. These smart sensing pixel arrays can generate dynamically changing images. Such a device is an optical-based sensor that can sense whether a target is moving in space, and is therefore also called a motion sensing sensor.

[0053] The present invention is based on the above principle and can realize the perception of moving targets through the motion sensing sensor according to the changes in light intensity caused by the movement of the object.

[0054] It's worth noting that the dynamic images captured by the motion-sensing sensor used in the present invention are not grayscale or color video images. This is because the dynamic images captured by the motion-sensing sensor used in the present invention only display binary or low-precision numerical changes (dynamic event clouds) in areas where light intensity changes. Areas where light intensity remains unchanged have no information output (thus, no background processing is required), and it can be assumed that the background has not changed. However, in video recordings, not only areas where light intensity changes are recorded and output, but also areas where light intensity remains unchanged. Video image processing systems must process images to distinguish useful moving objects. In contrast, the motion-sensing sensor used in the present invention does not require background processing, which not only reduces processing complexity but also ensures privacy in the image.

[0055] The optical motion pattern perception system disclosed in the present invention includes an image sensor, an image processing module, a motion detection module, and a timing monitor. The image sensor captures images in real time and generates an event based on the sensitivity of each pixel to local temporal light intensity changes. The events generated by all pixels of the image sensor together constitute a dynamic event cloud, and a binary or low-precision digital signal is generated based on the perception of a specific event cloud based on light intensity changes. The image processing module predefines a path on the image and predefines several specific areas on the path. A trigger pixel threshold is set for each specific area. When the number of pixels that change in a specific area exceeds its trigger pixel threshold, the specific area is triggered to form a specific event cloud. The motion detection module includes at least one state machine, which has N states, where N is a positive integer greater than 1. The state machine selects one state from the set of N states as the initial state. The state machine is controlled by the digital signal generated by the image sensor based on the specific event cloud, and generates multiple control signals based on the state of the state machine. The timing monitor can record the time of the specific event cloud in real time, is associated with the state machine of the motion detection module, and can return the state machine to the initial state. Compared with traditional recognition technologies such as infrared thermal sensing and wireless radio frequency identification, the system and method disclosed in the present invention can adapt to the needs of human-computer interaction developing towards artificial intelligence, which requires higher motion accuracy; compared with new recognition methods represented by face recognition that focus on software and light on hardware, the optical motion pattern perception method disclosed in the present invention not only does not require the extraction of basic information, but also does not require background processing, thus reducing the amount of data. It can generate binary or low-precision digital signals based on the perception of light intensity changes, greatly reducing the demand for computing power, thereby lowering the threshold for implementation, and ultimately enabling massive applications.

[0056] Example 1: An optical motion pattern sensing system

[0057] Example 1 provides an optical motion pattern sensing system, and its structure is described in detail below with reference to the accompanying drawings.

[0058] 1 , the optical motion pattern sensing system mainly includes four main parts, namely, an image sensor a, an image processing module b, and a motion detection module c. It may also include a timing monitor d, and the motion detection module c and the timing monitor d are connected.

[0059] Image sensor a captures image 100, and changes in light intensity trigger pixel events in its pixels. The output formed by all pixel events of image sensor a constitutes an event cloud.

[0060] The image processing module b is used to process the generated image 100, that is, to predefine a path 101 on the image 100 and to predefine a number of specific areas 102 on the path 101. A trigger pixel threshold is set for each specific area 102. When the number of pixels that change in a specific area 102 exceeds its trigger pixel threshold, the specific area 102 is triggered to form a specific event cloud.

[0061] A motion detection module c includes at least one state machine having N states, where N is a positive integer greater than 1. The state machine selects one state from a set of N states as an initial state. The state machine is controlled by a digital signal generated by the image sensor a in response to a specific event cloud, and generates a plurality of control signals based on the state of the state machine.

[0062] The timing monitor d can record the time of a specific event cloud in real time and is associated with the state machine, so as to enable the state machine to return to the initial state.

[0063] Among them, the image sensor a collects images 100 in real time and generates a dynamic event cloud;

[0064] When the event cloud moves along the path 101 predefined by the image processing module b on the image 100 and passes through the specific areas 102 in sequence in a time series, if the number of pixels changed in a specific area 102 exceeds the trigger pixel threshold, the specific area 102 is triggered to form a specific event cloud; the image sensor a senses the specific event cloud based on the change in light intensity and generates a binary or low-precision digital signal, which is then transmitted to the motion detection module c;

[0065] Once the state of the state machine encounters a deadlock or timeout phenomenon, the timing monitor d returns the state machine to the initial state according to the control signal generated by the motion detection module c or its own built-in timeout information;

[0066] Until the timing relationship between the specific event clouds triggered by these specific areas 102 meets the specific requirements of the event cloud in the specific area 102 of the path 101, the motion inspection module c generates multiple control signals according to the state that needs to change in the state machine to realize the state transition of the state machine.

[0067] Although the event cloud crosses these predefined specific areas in a predefined time sequence, the exact absolute time order of these crossings is not stable, but the relative logical time sequence is stable. The intersection of a predefined specific area on the image 100 can be determined by a simple trigger pixel threshold.

[0068] The motion detection module c detects motion patterns in a state machine. Each motion pattern detection is performed by a predefined token-based state machine. This state machine has N states, and the transitions between tokens are enabled by the intersection of specific event clouds over predefined regions. When a token reaches a specific state, a motion pattern is detected.

[0069] Among them, when the state machine encounters a deadlock phenomenon, an abnormal situation will occur. This abnormal situation can occur during the movement of the event cloud along the path 101, or it can occur when the event cloud ends its movement along the path 101. Specifically, the abnormal situation is caused by different reasons, including unexpected movement patterns, noise or light pollution, etc.

[0070] The state of the state machine encounters a timeout phenomenon, which includes not only the time gap between specific event clouds that occurs during the movement of the event cloud along the path 101 but also the time when the event cloud ends moving along the path 101 and the non-initial state continues to time out, and the state machine needs to return to the initial state.

[0071] As can be seen, regardless of whether an abnormality occurs during the movement of the event cloud along path 101, or even when the event cloud has completed its movement along path 101, if the state machine encounters a deadlock or timeout, timer monitor d will return the state machine to its initial state. For example, if the time interval between specific event clouds times out, timer monitor d will reset the state machine, returning it to its initial state. Another example of deadlock caused by unexpected movement patterns, noise, or light pollution can be eliminated by resetting the state machine using timer monitor d.

[0072] Specifically, the image sensor a is a motion image sensor, which includes a pixel matrix and a pixel control circuit, each pixel has a pixel structure including a photodetector and a storage unit, the pixel control circuit is capable of sequentially writing photodetection values ​​representing after exposure to the storage unit of the pixel, and the image sensor a further includes a comparison circuit, which is used to store the sequential photodetection values ​​of at least part of the pixels, to compare the sequential photodetection values, to generate a comparison signal based on the sequential photodetection values ​​of the pixels, and to generate a comparison matrix reflecting changes in the image 100 from the comparison signal.

[0073] It should be noted that, first, the motion image sensor is an image sensor with special functions, which has the basic function of acquiring images as an image sensor. Secondly, the motion image sensor has a pixel matrix and a pixel control circuit, each pixel of which is sensitive to changes in local temporal light intensity and generates an event. All pixels on the pixel matrix of the motion image sensor together constitute a dynamic event cloud based on all events generated by their sensitivity to changes in local temporal light intensity.

[0074] An event cloud is a collection of many events. If an event generated by a pixel is called a "raindrop," then the collection of events generated by many pixels together constitutes a "cloud." Regions within this "cloud" that are sensitive to changes in light intensity will experience dynamic changes, so the image 100 obtained by the motion image sensor is a dynamic image.

[0075] When the event cloud moves on the image 100 along the path 101 predefined by the image processing module b, it passes through a specific area 102 predefined on the path 101. If the number of pixels changed in a specific area 102 exceeds the trigger pixel threshold, the specific area 102 is triggered to form a specific event cloud.

[0076] Finally, since it has a comparison circuit that stores sequential photoelectric detection values ​​and compares sequential photoelectric detection values, it can generate a comparison signal based on the sequential photoelectric detection values ​​of the pixels, and generate a comparison matrix that reflects the changes in the image 100 from the comparison signal. Therefore, the motion image sensor can perceive specific events based on changes in light intensity and generate binary or low-precision digital signals, and transmit the digital signals to the motion detection module c. This enables the motion image sensor to have the ability to detect and compare calculations, and can detect changes in motion patterns.

[0077] Regarding motion image sensors, Shanghai Honggan Microelectronics Technology Co., Ltd. has applied for a patent. For more specific structures, please refer to the Chinese invention patent "Improved CMOS Image Sensor" with application number 202210132353.1, which will not be repeated here.

[0078] Specifically, the timing monitor d is a watchdog timer, whose English name is Watchdog timer.

[0079] Specifically, the image processing module b and the motion detection module c can be in either hardware or software form.

[0080] The image processing module b and the motion detection module c are respectively connected to the image sensor a.

[0081] As a specific implementation, the image processing module b and the motion detection module c are integrated on the image sensor a to form an image sensor a that is motion-sensitive and intelligent, and together with a timing monitor d associated with a motion / time-based state machine, constitute the two major hardware blocks of the optical motion pattern perception system.

[0082] As another specific embodiment, the image sensor a, image processing module b, motion detection module c, and timing monitor d can be independent of each other. When these four components are independent, the optical motion pattern perception system preferably also includes a controller. The controller can be specifically configured to determine whether the timing relationship between the specific event clouds triggered by these specific regions meets the specific requirements of the event clouds in the specific region 102 of the path 101. The image sensor a, image processing module b, motion detection module c, and timing monitor d are each connected to the controller. In this case, the image sensor a acts more like a camera, capable of sensitively generating dynamic images based on local temporal changes in light intensity. The computational processing and determination functions are delegated to the controller.

[0083] The optical motion pattern sensing system disclosed in the present invention can be further applied. For example, to obtain a wider field of view, a common embodiment is that the optical motion pattern sensing system includes multiple image sensors a, and the multiple image sensors a are arranged in a manner to maximize the field of view of motion sensing.

[0084] The plurality of image sensors a are connected to the controller in sequence or respectively;

[0085] Each image sensor a is configured with an image processing module b. Each image sensor a predefines a path 101 on the image 100 captured by the corresponding image sensor a to form multiple paths 101, and predefines several specific areas 102 on each path 101. The several specific areas 102 of the multiple paths 101 are associated with one or more state machines.

[0086] In order to obtain a larger field of view, another specific implementation is to adjust the field of view angle of the image sensor a so that the image sensor a has a large field of view. The optical motion pattern perception system includes multiple image processing modules b, and the multiple image processing modules b are respectively connected to one image sensor a or respectively connected to the controller;

[0087] The plurality of image processing modules b predefine paths 101 on an image 100 captured by an image sensor a to form a plurality of paths 101 , and predefine a plurality of specific areas 102 on each path 101 , as shown in FIG2 ;

[0088] Several specific regions 102 of the plurality of paths 101 are associated with one or more state machines.

[0089] The above specific implementations of obtaining a larger field of view can all realize multiple paths 101 for detecting intrusion, and are applicable to the scenario of application example 3. Regardless of the method, it is within the protection scope of the optical motion pattern perception system disclosed in the present invention.

[0090] Example 2: An optical motion pattern perception method

[0091] Example 2 provides an optical motion pattern sensing method, using the optical motion pattern sensing system of Example 1, the method comprising the following steps:

[0092] Dynamic image acquisition: Image sensor a acquires image 100 in real time. Changes in light intensity trigger pixel events within its pixels. The output formed by all pixel events of image sensor a constitutes an event cloud.

[0093] Image processing: The image processing module b predefines a path 101 on the image 100 and predefines a number of specific areas 102 on the path 101;

[0094] Generation and perception of a specific event cloud: The event cloud moves along a path 101 on the image 100 and passes through specific regions 102 in a temporal sequence. If the number of pixels in a specific region 102 that changes exceeds its trigger pixel threshold, the specific region 102 is triggered to form a specific event cloud.

[0095] Image sensor a senses the specific event cloud based on light intensity changes and generates a binary or low-precision digital signal. The digital signal is then transmitted to motion detection module c and timing monitor d. The occurrence time of the specific event cloud is recorded in real time by timing monitor d.

[0096] State transition of the state machine: until the timing relationship between the specific event clouds triggered by these specific areas meets the specific requirements of the event clouds in the specific areas of the path, the motion verification module c generates multiple control signals according to the state of the state machine that needs to change, thereby realizing the state transition of the state machine;

[0097] Wherein, once the state machine encounters a deadlock or timeout phenomenon, the timing monitor d returns the state machine to the initial state according to the control signal or timeout information generated by the motion inspection module c.

[0098] Specifically, when the state machine encounters a deadlock phenomenon, an abnormal situation will occur. This abnormal situation may occur during the movement of the event cloud along the path 101, or may occur at the end of the movement of the event cloud along the path 101. Specifically, the abnormal situation is caused by different reasons, including unexpected movement patterns, noise or light pollution, etc.

[0099] The state of the state machine encounters a timeout phenomenon, which includes not only the time gap timeout between specific event clouds during the movement of the event cloud along the path 101, but also the time timeout when the event cloud ends moving along the path 101. Regardless of the time timeout mode, the state machine needs to return to the initial state.

[0100] The question of "whether the timing relationships between the specific event clouds triggered in these specific areas meet the specific requirements of the event clouds in the specific areas of the path, thereby achieving the state transition of the state machine" needs to be determined based on the specific application scenario. The following three specific application examples of the optical motion pattern sensing method for waving switches, automatic doors, and window intrusion alarms further illustrate how the present invention uses the disclosed optical motion pattern sensing method to achieve state transitions in the state machine.

[0101] Application Example 1: Wave Switch

[0102] Inductive switches are often encountered in life. A wave switch is a common inductive switch. The following takes a simple wave switch as an example.

[0103] Referring to Figure 2, first, according to the needs of the perceived person, for example, the hand waving from the left to the right can be defined as "open", and the hand waving from the right to the left can be defined as "closed". The path 101 can be predefined as the perceived person's hand waving from the left to the right or from the right to the left, and two specific areas 102 can be predefined on the predefined path 101: a first specific area A1 and a first specific area B1, as shown in Figure 2 (Ⅰ).

[0104] Correspondingly, the state machine diagram can be defined as follows: if the first specific area A1 and the first specific area B1 are triggered in sequence according to the time sequence, forming the first specific event cloud ZA1 and the second specific event cloud ZB1 respectively, then starting from the initial state, the first specific event cloud ZA1 is generated first, and then the second specific event cloud ZB1 is generated, then the state machine generates a state - "open"; conversely, starting from the initial state, the second specific event cloud ZB1 is generated first, and then the first specific event cloud ZA1 is generated, then the state machine generates another state - "closed", as shown in Figure 2(Ⅱ).

[0105] Based on the above analysis of application example 1, it can be summarized as follows:

[0106] The predefined path is: a straight line or an arc formed by the hand of the subject waving from the left to the right or from the right to the left;

[0107] The predefined specific areas are: a first specific area A1 and a first specific area B1 are predefined on the left and right sides of the path 101 respectively;

[0108] The timing relationship between the specific event clouds triggered by these specific areas meets the specific requirements of the event cloud in the specific areas of the path, and realizes the state conversion of the state machine: the event cloud passes through the first specific area A1 and the first specific area B1 along the path 101. If the first specific area A1 and the first specific area B1 are triggered in sequence according to the chronological order, forming the first specific event cloud ZA1 and the second specific event cloud ZB1 respectively, then starting from the initial state, the first specific event cloud ZA1 is generated first, and then the second specific event cloud ZB1 is generated, then the state machine generates a state; conversely, starting from the initial state, the second specific event cloud ZB1 is generated first, and then the first specific event cloud ZA1 is generated, then the state machine generates another state.

[0109] Application Example 2: Automatic Door

[0110] As a contactless door opening method, the automatic door is very useful for the physical contact between people and the environment. A very simple automatic door can be designed using the present invention.

[0111] Referring to Figure 3, first, based on the fact that the person being sensed is outside the door, the person's feet walk from outside to inside in chronological order, and three specific areas 102 can be predefined in front of the door: the third specific area A2, the fourth specific area B2 and the fifth specific area C2, as shown in Figure 3 (Ⅰ).

[0112] Correspondingly, the state machine diagram can be defined as: starting from the initial state, the third specific event cloud ZA2 is generated first, then the fourth specific event cloud ZB2 is generated, and finally the fifth specific event cloud ZC2 is generated, then the state of the state machine changes to "open the door", as shown in Figure 3(Ⅱ).

[0113] Based on the above analysis of Application Example 2, we can conclude that:

[0114] The predefined path is: the route that the perceived person's feet walk from outside the door to inside the door;

[0115] The predefined specific areas are: according to the chronological order of the foot walking from the outside to the inside, a third specific area A2, a fourth specific area B2 and a fifth specific area C2 are predefined from the outside to the inside on the path 101;

[0116] The timing relationship between the specific event clouds triggered by these specific areas meets the specific requirements of the event cloud in the specific areas of the path, realizing the state conversion of the state machine: the event cloud passes through the third specific area A2, the fourth specific area B2 and the fifth specific area C2 along the path 101. If the third specific area A2, the fourth specific area B2 and the fifth specific area C2 are triggered in sequence according to the chronological order, forming the third specific event cloud ZA2, the fourth specific event cloud ZB2 and the fifth specific event cloud ZC2 respectively, then starting from the initial state, the third specific event cloud ZA2 is generated first, then the fourth specific event cloud ZB2 is generated, and finally the fifth specific event cloud ZC2 is generated, and the state of the state machine is converted.

[0117] Application Example 3: Window Break-in Alarm

[0118] Similar to the previous application example, the intrusion movement is characterized by extending from the outside to the inside through a specific path 101, such as through a window or an unlocked door. Therefore, the movement pattern from the outside to the inside can be detected and a high probability intrusion can be defined.

[0119] Referring to FIG4 , first, based on the time sequence in which the perceived person passes by the window from near to far after climbing over the window, three specific areas 102 can be defined within the window: the sixth specific area C3, the seventh specific area B3, and the eighth specific area A3, as shown in FIG4 (I).

[0120] Correspondingly, the state machine diagram can be defined as: starting from the initial state, the sixth specific event cloud ZC3 is generated first, then the seventh specific event cloud ZB3 is generated, and finally the eighth specific event cloud ZA3 is generated, then the state of the state machine changes to "alarm", as shown in Figure 4 (Ⅱ).

[0121] Based on the above analysis of application example 3, we can conclude that:

[0122] The predefined path is: the route that the perceived person takes from outside the window through the window to inside the window;

[0123] The predefined specific areas are: a sixth specific area C3, a seventh specific area B3, and an eighth specific area A3 are predefined on the path 101 in order of time after the perceived person passes through the window from outside to inside the window and moves away from the window;

[0124] The timing relationship between the specific event clouds triggered by these specific areas meets the specific requirements of the event cloud in the specific areas of the path, realizing the state conversion of the state machine: the event cloud passes through the sixth specific area C3, the seventh specific area B3 and the eighth specific area A3 along the path 101. If the sixth specific area C3, the seventh specific area B3 and the eighth specific area A3 are triggered in sequence according to the chronological order, forming the sixth specific event cloud ZC3, the seventh specific event cloud ZB3 and the eighth specific event cloud ZA3 respectively, then starting from the initial state, the sixth specific event cloud ZC3 is generated first, then the seventh specific event cloud ZB3 is generated, and finally the eighth specific event cloud ZA3 is generated, and the state of the state machine is converted.

[0125] However, in practice, since a building has more than one window, in order to monitor multiple windows of a building, it is necessary to monitor multiple windows simultaneously, which requires a wider field of view to monitor multiple intrusion paths 101.

[0126] Furthermore, how to obtain a larger field of view? One way is to increase the number of image sensors a, and multiple image sensors a are arranged in a manner to maximize the field of view of motion perception; another way is to increase the number of pixel matrices within the image sensor a, so that a single image sensor a obtains a large field of view, and then increase the number of image processing modules b. For example, prisons need to monitor prisoners escaping from multiple paths, or certain important places such as banks and vaults need to prevent thieves from invading from multiple entrances, which requires a larger monitoring field of view and simultaneous monitoring of multiple paths 101.

[0127] Whether entering through a door or a window, Application Examples 2 and 3 are similar. The "temporal relationships between specific event clouds triggered in specific areas satisfy the specific requirements of the event clouds in specific areas of the path, thereby achieving state transitions in the state machine" in these two application examples can be summarized as follows:

[0128] The event cloud passes through three specific areas along path 101. If the three specific areas are triggered in chronological order and form three specific event clouds respectively, then starting from the initial state, the three specific event clouds are generated in sequence, and the state of the state machine is transformed, as shown in Figures 3 and 4.

[0129] It can be seen from this that the optical motion pattern perception method provided by Example 2 has a wide range of applications.

[0130] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. An optical motion pattern sensing system, characterized in that: include At least one image sensor (a) collects images (100) in real time, and triggers pixel events in its pixels due to changes in light intensity, and the output formed by all pixel events of the image sensor (a) constitutes an event cloud; At least one image processing module (b) predefines a path (101) on the image (100), and predefines a plurality of specific areas (102) on the path (101), wherein a trigger pixel threshold is set for each specific area (102), and when the number of pixels changed in a specific area (102) exceeds its trigger pixel threshold, the specific area (102) is triggered to form a specific event cloud; at least one motion detection module (c), comprising at least one state machine, the state machine having N states, wherein N is a positive integer greater than 1, the state machine selecting a state from a set of N states as an initial state, the state machine being controlled by a digital signal generated by the image sensor (a) according to a specific event cloud, and generating a plurality of control signals according to the state of the state machine; The image sensor (a) collects images (100) in real time to generate a dynamic event cloud; the event cloud moves on the image (100) along a path (101) predefined by the image processing module (b), and passes through the specific areas (102) in sequence according to a time sequence. If the number of pixels changed in a specific area (102) exceeds the trigger pixel threshold, the specific area (102) is triggered to form a specific event cloud; the image sensor (a) senses the specific event cloud according to the change in light intensity to generate a binary or low-precision digital signal, and transmits the digital signal to the motion detection module (c); until the timing relationship between the specific event clouds triggered by these specific areas meets the specific requirements of the event cloud in the specific area (102) of the path (101), the motion detection module (c) generates a plurality of control signals according to the state of the state machine that needs to change, so as to realize the state transition of the state machine.

2. The optical motion pattern sensing system according to claim 1, characterized in that: It also includes a timing monitor (d), which can record the time of a specific event cloud in real time and is associated with the state machine of the motion inspection module (c) to return the state machine to an initial state; Once the state of the state machine encounters a deadlock phenomenon or a timeout phenomenon, the timing monitor (d) returns the state machine to an initial state according to the control signal or timeout information generated by the motion inspection module (c).

3. The optical motion pattern sensing system according to claim 1 or 2, characterized in that: The image sensor (a) is a motion image sensor, which includes a pixel matrix and a pixel control circuit, each pixel has a pixel structure including a photodetector and a storage unit, the pixel control circuit is capable of sequentially writing a photodetection value representing a photodetection value after exposure to the storage unit of the pixel, and the image sensor (a) also includes a comparison circuit, the comparison circuit is used to store the sequential photodetection values ​​of at least part of the pixels, and is used to compare the sequential photodetection values ​​of the pixels. The method comprises the steps of: generating a comparison matrix for reflecting image changes from the comparison signal and generating a comparison matrix for reflecting image changes from the comparison signal.

4. The optical motion pattern sensing system according to claim 3, characterized in that: The image processing module (b) and the motion detection module (c) are integrated on the image sensor (a).

5. The optical motion pattern sensing system according to any one of claims 1 to 3, characterized in that: The system further comprises a controller, wherein the image sensor (a), the image processing module (b), the motion detection module (c) and the timing monitor (d) are independent of each other and are respectively connected to the controller.

6. The optical motion pattern sensing system according to claim 3 or 5, characterized in that: The device comprises a plurality of image sensors (a), wherein the plurality of image sensors (a) are arranged in a manner to maximize the field of view for sensing motion; The plurality of image sensors (a) are connected to the controller in sequence or respectively; Each image sensor (a) is equipped with an image processing module (b), each image sensor (a) predefines a path (101) on an image (100) captured by the corresponding image sensor (a) to form a plurality of paths (101), and predefines a plurality of specific areas (102) on each path (101), Certain areas (102) of the plurality of paths (101) are associated with one or more state machines.

7. The optical motion pattern sensing system according to claim 3 or 5, characterized in that: comprising a plurality of image processing modules (b), wherein the plurality of image processing modules (b) are respectively connected to the image sensor (a) or respectively connected to the controller; A plurality of image processing modules (b) predefine paths (101) on an image (100) captured by an image sensor (a) to form a plurality of paths (101), and predefine a plurality of specific areas (102) on each path (101). Certain areas (102) of the plurality of paths (101) are associated with one or more state machines.

8. An optical motion pattern sensing method, characterized in that: include Dynamic image acquisition: The image sensor (a) acquires images (100) in real time, and changes in light intensity trigger pixel events in its pixels. The output formed by all pixel events of the image sensor (a) constitutes an event cloud; Image processing: the image processing module (b) predefines a path (101) on the image (100), and predefines a number of specific areas (102) on the path (101); Generation and perception of a specific event cloud: the event cloud moves along a path (101) on an image (100) and passes through the specific areas in sequence according to a time series. If the number of pixels changed in a specific area (102) exceeds its trigger pixel threshold, the specific area (102) is triggered to form a specific event cloud; the image sensor (a) senses the specific event cloud according to the change in light intensity to generate a binary or low-precision digital signal, and transmits the digital signal to the image sensor (a). Sent to the motion inspection module (c); State transition of the state machine: until the timing relationship between the specific event clouds triggered by these specific areas meets the specific requirements of the event cloud in the specific area (102) of the path (101), the motion inspection module (c) generates multiple control signals according to the state of the state machine that needs to change, thereby realizing the state transition of the state machine.

9. The optical motion pattern sensing method according to claim 8, characterized in that: When the image sensor (a) senses the specific event cloud according to the change of light intensity, generates a binary or low-precision digital signal, and transmits the digital signal to the motion detection module (c), the image sensor (a) also transmits the digital signal to the timing monitor (d), and the occurrence time of the specific event cloud is recorded in real time by the timing monitor (d); And once the state of the state machine encounters a deadlock phenomenon or a timeout phenomenon, the timing monitor (d) returns the state machine to an initial state according to the control signal or timeout information generated by the motion inspection module (c).

10. The optical motion pattern sensing method according to claim 9, characterized in that: The predefined path is: a straight line or an arc when the hand of the subject waves from the left to the right or from the right to the left; The predefined several specific areas are: a first specific area (A1) and a second specific area (B1) are respectively predefined on the left and right sides of the path (101); The timing relationship between the specific event clouds triggered by these specific areas meets the specific requirements of the event cloud in the specific areas of the path, and the state conversion of the state machine is realized as follows: the event cloud passes through the first specific area (A1) and the first specific area (B1) along the path (101). If the first specific area (A1) and the first specific area (B1) are triggered in sequence according to the chronological order, forming the first specific event cloud (ZA1) and the second specific event cloud (ZB1) respectively, then starting from the initial state, the first specific event cloud (ZA1) is generated first, and then the second specific event cloud (ZB1) is generated, and the state machine generates one state; conversely, starting from the initial state, the second specific event cloud (ZB1) is generated first, and then the first specific event cloud (ZA1) is generated, and then the state machine generates another state.

11. The optical motion pattern sensing method according to claim 9, characterized in that: The predefined path is: the route that the perceived person's feet walk from outside the door to inside the door in sequence; The predefined several specific areas are: according to the foot walking from the outside to the inside in chronological order, a third specific area (A2), a fourth specific area (B2) and a fifth specific area (C2) are predefined from the outside to the inside on the path (101); The temporal relationship between the specific event clouds triggered by these specific areas meets the specific requirements of the event cloud in the specific area of ​​the path, and the state conversion of the state machine is realized: the event cloud passes through the third specific area (A2), the fourth specific area (B2) and the fifth specific area (C2) along the path (101). If the third specific area (A2), the fourth specific area (B2) and the fifth specific area (C2) are triggered in chronological order, they form The third specific event cloud (ZA2), the fourth specific event cloud (ZB2) and the fifth specific event cloud (ZC2) are generated. Then, starting from the initial state, the third specific event cloud (ZA2) is generated first, then the fourth specific event cloud (ZB2), and finally the fifth specific event cloud (ZC2), and the state of the state machine is converted.

12. The optical motion pattern sensing method according to claim 9, characterized in that: The predefined path is: the route that the perceived person walks through from outside the window to inside the window; The predefined specific areas are: a sixth specific area (C3), a seventh specific area (B3) and an eighth specific area (A3) are predefined in sequence on the path (101) after the perceived person passes through the window from outside the window and enters the window, and then moves away from the window in chronological order; The timing relationship between the specific event clouds triggered by these specific areas meets the specific requirements of the event cloud in the specific areas of the path, and the state conversion of the state machine is realized as follows: the event cloud passes through the sixth specific area (C3), the seventh specific area (B3) and the eighth specific area (A3) along the path (101). If the sixth specific area (C3), the seventh specific area (B3) and the eighth specific area (A3) are triggered in chronological order, forming the sixth specific event cloud (ZC3), the seventh specific event cloud (ZB3) and the eighth specific event cloud (ZA3) respectively, then starting from the initial state, the sixth specific event cloud (ZC3) is generated first, then the seventh specific event cloud (ZB3) is generated, and finally the eighth specific event cloud (ZA3) is generated, and the state of the state machine is converted.

13. Application of the optical motion pattern sensing method according to claim 8 or 9 in hand-wave switches, automatic doors or window intrusion alarms.

Images