SE2030138A1 - Personal detector for detecting people on a property's land or plot, or in a building - Google Patents

Abstract

The present invention is a camera-based detector that searches for and detects selected objects moving in an environment. The detection takes place by comparing digital images in an image stream against a reference image and the changes that are detected are then compared with shapes that represent the objects that are searched for. The reference image is generated continuously by weighing previous images. When the camera detector detects an object and verifies that it is a searched object, a detector signal is created which can trigger an external alarm directly or forwarded to a receiving unit for further handling. The camera detector can be a complete detector for detecting unauthorized persons moving as well. indoors and outdoors. The camera detector can monitor the ground or plot of a property with detection when someone crosses the border or moves over an area, or monitor areas inside houses or buildings and generate a detector signal when a person moves within the monitored area. The camera detector can effectively distinguish between humans and animals, between adults and children, or between people and moving objects.

Description

Image-based detector unit with efficient personal detection in a property as well as on a property's plot or land.

Technical Field The present invention relates to a product which can detect and alarm when an object, such as a person or a vehicle, enters or moves over a property or land of a property, or when a person moves on or in a property.

The detection takes place by an image-based detector detecting an object that is moving and verifying if it is a permitted or unauthorized object, and if it is not permitted, a signal to the connected alarm or monitoring unit indicates that a detection has taken place.

The function is in one design to detect personal intrusion on a property already close to the boundary of the property and thus provide increased ability to warn early before a possible burglary or attempted theft, and in another design, other intrusions or different types of objects can be detected, such as vehicle or person movements across town -, regional, or land borders, or relocations within an area with no access for unauthorized persons (eg energy or defense installations).

State of the art There are today a variety of solutions and products for detecting unauthorized persons or vehicles in connection with various areas or buildings being monitored and they are generally based on detecting physical or material damage, such as opening a door, smashing a window or a signal from a PIR detector, after which a manual examination of the cause is performed visually.

Difficulties for PIR detectors to distinguish between people and animals, especially when animals over 15 kg, means that properties and buildings where there is a risk of this type of waste detection usually have to refrain from using PIR detection and thus reduces the ability of surveillance to detect if unauthorized persons touch in or on property.

Different types of visual surveillance are based on manual camera surveillance where people such as the property owner or people in alarm centers and other places visually monitor and assess events based on what they see on their screens and monitors. For special purposes such as police or security services to receive multiple digital videos from connected camcorders and analyze them in real time to find people and faces.

For outdoor surveillance, there are currently no solutions available in addition to traditional camera surveillance and there are no detectors for outdoor use other than PIR detectors that can detect an intrusion on a property's area or plot. Although there are battery-powered video cameras for outdoor use with a wireless connection, these are based on a PIR detector starting them up and they completely lack the ability to automatically distinguish between, for example, animals and humans. There are camcorders for outdoor use for continuous manual video surveillance, but this type of camcorder requires a network connection for continuous operation.

There is currently no known system for automated monitoring of the personal movements of private properties, private residences or company properties in a building, in a property or on a property's plot or land, where the monitoring takes place with automated detection of personal intrusion and where automated personal detection outdoors excludes battery-powered equipment camera detectors or other battery-powered personal detectors with continuous monitoring, both with and without personal verification. Furthermore, there are currently no camera-based detectors that co-battery operation and wireless connection can automatically detect and verify personal movements both indoors and outdoors, at least not for months or years in a row without battery replacement.

The following patents discuss camera or video-based detection of objects or shapes: US2012201470 (A1) is a method of finding and finding all the different contours in a digital image by finding differences between nearby pixel values and then managing the areas that these contours form.

US8682071B1 is a method of creating sketch-like images of various objects based on a digital photo of the object.

CN110008927 is a method to help the police surveillance center to video monitor areas with the help of video cameras by real-time and with advanced image analysis of digital film analyze people's movements and how people move in relation to them in any way acting or interacting with each other, risk assessment and an alert signal. is generated when certain predetermined movements or movement patterns are detected, so that police surveillance personnel can focus their attention on the monitor or monitors in question. US200758837A1 is a method of comparing two digital color images by dividing the images into squares and then measuring the total intensity of each color in the single image after which they are compared with the corresponding value in the corresponding box of the other image.

Problem Solving The present invention solves the problems of the prior art in that it has the features set forth in the appended claims.

The present invention is an image-based detector that searches for specific objects by continuously comparing and finding changes in a stream of images. When a change between current and previous images is detected, it is checked that it corresponds to one or more of these shapes.

Object detection in accordance with the present invention may in one design cause a person to be detected in an environment persons may be expected to be as in a building or on a property plot or land, and in another design in an environment where persons are not expected to be as in a security classified area with said object detection may in one application refer to detection when a person or a vehicle enters a property or land on a property, in another application means detection when a person moves on or in a property or area, in a third application detect when a person or a vehicle crosses a boundary such as a solitary boundary, a property boundary or a land boundary, in a fourth application it detects a person who is in water such as in a pool, pool or out at sea or sea and detects if the person for some reason falls below the water surface with alarm and drowning accidents. accident risk the time under the water surface becomes too long, for the prevention of The present invention handles a stream of digital images in a very simple and energy efficient way, with a limited number of pixels per image and a limited number of images per second low energy consumption is obtained which leads to increased possibilities for continuous automated monitoring communication and battery operation.

BRIEF DESCRIPTION OF THE DRAWINGS Figure 1: An embodiment of the present invention.

Figure 2: Example of how two images of an object in motion can give completely different areas of pre-detection.

Figure 3: Flow chart for a design of object detection.

Figure 4: Flow chart for a design of object detection in connection with object movements between subsequent images.

Detailed Description of the Invention An image-based detector for detecting one or more objects entering, moving over or within, or leaving an area under surveillance. People, animals or vehicles that enter, move over, or leave an area are examples of objects you want to discover, and said area can be a property's plot or land, a single-building exterior or interior, or any other type of area where you want to detect some objects appear or are in motion.

Said detector will henceforth be called video camera detector.

A camcorder detector may in one design be intended for mains operation, in another design be intended for battery operation, and in another design be intended for both mains operation and battery operation.

The camcorder detector is based on a car recording unit (11) for capturing series of digital images of an area which may be a monitored area or of another area of interest for the detection of predetermined objects, a computer unit processing images (12), and a communication unit (13). ) for communication with a receiving device.

Said camcorder detector may in one embodiment be a composite unit consisting of said image pickup unit, said computer unit and said communication unit, and in another design may consist of a composite unit consisting of said image capture unit and said communication unit while said computer unit is located in an external image receiving unit.

Said image capture unit may in one design consist of an image sensor for image capture through a lens, in another embodiment be based on or consist of a digital video camera, in another design be based on or consist of a digital still camera with photography in series of images, and in another design be a combination of one or more image sensors, one or more digital camcorders, one or more digital still cameras. Said image pickup units may be intended for the visible light spectrum, for the infrared (IR) spectrum, or for a combination of both the visible spectrum and the IR spectrum. Said image capture unit can store images in different designs based on the intensity of IR radiation, the intensity of single color or the intensity of the image spectrum (grayscale), which means that subsequent image processing and calculations can be reduced to handling only one value per pixel. Said image capture unit can in another design store images in color with more than one value per pixel.

Said digital video camera can in one design switch mode of operation between digital film with 30 frames per second (FPS) or more, and series of images with less than 30 FPS. With fewer FPS, either image capture can be slowed down or the stream of images can sample the flow of images from an image flow with more FPS.

Said image sensor can in one design switch mode of operation between digital film with a larger number of images per second (FPS), and image series with fewer FPS. With fewer images per second, the shooting can be adjusted to a greater extent as needed, either by slowing down the shooting or the stream of images can be sampled from the flow of images from said digital film, from a few images per second to digital film with 30 FPS or more.

A communication unit for wireless and / or wired communication with another camcorder detector may in one design be equipped with a unit and said other unit may in one design consist of a central unit for surveillance, in another design consist of a receiving unit at an alarm center, another design enable communication of external signaling equipment such as sound, light. Said communication may in one design be network communication which may be wired or wireless such as Wifi, in another design be some other type of radio communication, and in another design be with sound, light or other form of communication . Said central unit may in one design consist of a unit of false alarm, in another design of a monitoring unit for monitoring a single property or a facility. Said plant may in one design consist of an amusement plant such as a bathing plant or a leisure plant, in another design of a technical plant such as a power plant or a track area for railway or tram, and in a fourth design of some other type of plant.

A camcorder detector may in one design be equipped with separate units for image capture and computer unit with software for the image processing and calculations required for detecting said object, in another design be equipped with digital video or still image camera connected to a computer unit with said video or software, in another design be equipped with a digital still camera with its own capacity for said software and where said software is included in said video or still camera.

Said detection is based on said object moving and these movements cause images in an image sequence to change, and said detection takes place through image comparisons where differences between images are analyzed and compared with the previous image or images, and when differences are discovered these differences are compared with the shapes or figures said object, and when conformity is found it is considered that said object has been found.

Said image comparisons can in different formats handle images based on one or more values per pixel, and if said image is an image with more than one value per pixel such as RGB color image, said image comparison can take place on one or more of these different values per pixel. For example, said image comparison can be made based on only the green pixel value, which means that subsequent image processing and calculations can be simplified to handle only one value per pixel even for an RGB color image.

Said image comparisons can be performed with different techniques to detect image changes obtained when said object moves. Said detection of image changes can be performed with known methods for motion detection such as Temporal Frame Differencing, Optical Flow or Background Subtraction.

The present invention has no requirements as to the technique used and the variation of Background Subtraction described herein for illustrating how a detection can be performed. Said background subtraction means that a reference background is subtracted away from the current image and what remains represents an image of a movement. Said reference background requires a learning period to be generated and since the background changes over time as details are added or lost, details move such as a bush's movements in the wind, light changes with day and season variations, or in connection with headlights or outdoor lighting, said reference background must also be updated continuously. Board reference background must therefore change over time and this is done by in the image reference background, in another design calculate its pixels by weighing the simplest design for static environments using the previous as each individual pixel value from several previous images. Said aggregation can be linear with equal weight for each pixel in the different images, or non-linear with decreasing value for pixels of older images.

Said image of a movement can then be handled in different ways so that a searched object can be found and the search for said object can be done by comparing said image in one design with stored shapes of said object, in another design with stored pixels or values in a matrix. The search for objects can also be done by transforming said image of a movement in a design through calculations into stored values in matrix or vector form whereupon said applicant by comparison with shapes stored in the same way, or in another design comparisons between some other type of stored digital values or digital representation of both said image of a movement and said shapes.

Method for finding said object in said software can be illustrated in a design according to figure 3. Compare connected areas with stored shapes to detect, scale up or down said stored shapes for better matching, if necessary divide stored shapes into smaller parts for safer matching , and when sufficient conformity is found, this is considered as the said object being found.

Coherent areas can in a design be stored in matrices or vectors with pixel row values or pixel group values, for comparisons and detection of objects in accordance with the method described above.

Coherent areas can in another design be assessed based on the number of pixel or matrix elements, possibly weighed with mutual distances, set in relation to any detection level set based on the shapes to be detected.

Contiguous areas, another design can be stored in a bit-based matrix with individual pixel values represented by individual bits for false or true in a single processor word consisting of a number of bits such as one or more bytes of 8 bits, thus the search algorithm for comparing said contiguous areas and said stored forms are performed at the word level with true or false, and thus even comparisons in larger areas can be made extremely simple and extremely fast. Example: A 16-bit processor has the word length 16 bits and each pixel within the contiguous area is represented by a one, thus a word can represent 16 pixels and each word level comparison gives the answer true or false, and since each word indicates a value corresponding to the number of ones in the word , when subtracting the answer, the false bit can be supplemented by the number of bits that differ, and this number of bits that differ can form the basis for assessing whether the difference is large enough for the comparison to be considered false, or if the difference is so small that it can be considered negligible. .

Objects to be found are represented by one or more stored shapes, and these can in one design be predefined and stored as object shapes linked to standard formats to look for, in another design are defined or received as additional or changed object shapes over time, and in another design consist of a combination of predefined and additional or changed object shapes.

Said shapes may consist of stored pixels or values in a matrix, of stored pixels or values in a vector, or of some other type of digital values or digital representation of said shapes.

When it can be established that the said image is in sufficient accordance with a search object, we have received a detection, and if necessary, a subsequent phase of de-verification can then be started.

Said verification can take place in a similar manner as described above, but must also take into account the problems that arise when an object moves and cause changes in shape or size between images. Said problem is illustrated in figure 2 with two images (21, 22) of an object in movement to the right. Image (22) shows both the previous (dashed) and current (solid) location of the object, with two areas (24, 25) not included in both images, while a center area (23) is included in the boat images. The result of a comparison between the images is that the separate areas (24,25) are both detected as changes compared to the previous image, one (24) that one half of the object (26) disappears and the other (25) that half an object (26) is added The problem is exacerbated when one has to take into account movements towards the camera detector where the object grows as it approaches the camcorder detector, or shrinks the approach moves from the camcorder detector, or when it moves at some other angle to the camcorder detector.

Said problems can be solved in one design by adding alternative forms such as search objects and these alternative forms can be independent forms to search for or forms linked to a specific object, for example a car form can be linked to shorter, longer or higher forms and still represent a car .

Said problem can be solved in another design by taking into account in said software the image processing and calculations the said comparisons with stored forms for detection, and said consideration can in one design be based on said object being followed between images, and in another design in that said object is allowed to change shape between images, and in a third design is based on a combination of following an object and allowing shape changes. with the desired shape.

Said problem can be solved in another design by a combination of said alternative forms and said image processing and calculations.

Flow diagram for handling objects in motion in said software can in one design be illustrated according to figure 4. complete image receiving unit for further analysis before detection is confirmed and results in a an external device pre-monitoring resulting in some kind of increased vigilance or increased sensitivity for detection of movements or intrusions, in another design may be an alarm signal to the enlarger operator, and in another design may be a local alarm, or some combination of these or other measures.

Said camcorder detector can be functional for detection even under-weak light conditions by in one design being connected to be able to turn on single-area or object lighting which can be LED-based with visible light or IR radiation, and in another design be connected to or have built-in a camera flash with visible light or IR radiation.

In connection with dim lighting, fog or darkness, the video camera detector can be started up or triggered for detection even under low light conditions by an internal or external PIR detector or by some internal or external signal.

Said camcorder detector may in one design consist of a complete unit for detecting various objects and in another design of an image receiving unit for transmitting sequences of digital images or of a continuous stream of digital images to an external unit. Said image receiving unit may in one design consist of a single image sensor connected to a communication unit for said transmission, in another design of traditional digital video or still image cameras with the possibility of said transmission or connected to a communication unit for said transmission, and in another design of a digital video camera which can switch between different numbers of frames per second (FPS) with the possibility of said transmission or connected to a communication unit for said transmission.

For the detection of objects such as persons or vehicles, the resolution requirements are controlled by the distance from the camcorder detector to the object in question, and with lower resolution requirements the power consumption is kept down while maintaining good ability to detect, to the benefit of battery operation. By reducing the frame rate FPS, the load on the camcorder detector can also be reduced with reduced power consumption as a result, and with as low as 4 FPS, all types of personal movements of interest can be detected with high security (except possibly free fall), also for battery operation.

A camcorder detector can be equipped with a variable image range where part of the image is removed by removing pixel rows, pixel columns or blocks of pixels from handling, for example the sky above the trees can be removed, and with fewer pixels to handle reduces power consumption, for battery operation .

A camcorder detector can be equipped with variable image resolution where the number of pixels can be reduced by skipping pixels, for example where other pixels can be skipped for reduced sensitivity for long-distance objects or too small objects.

A camcorder detector can be equipped with variable frame rate (FPS) pre-adaptation to different objects different sizes and different speeds in object movements, small objects or fast movements require higher FPS while large objects or slow motion can allow lower FPS, and with lower FPS reduces power consumption, for battery operation. By said small and large objects, respectively, is meant their relative size of the image in question, not necessarily their size in meters.

A camcorder detector can be equipped with machine learning which in one design can enable the reception of new or updated shapes to detect, and in another design can enable the reception of image boundaries to reduce the number of pixels by removing eg houses, sky and other areas without interest in detection, or remove said image delimiters. A camcorder detector may be equipped with machine learning for automated setting or remote setting of image resolution, frame rate, image delimitations or other image properties, enabling in one design automated image resolution setting based on data sent from an external device such as a central unit, alarm or alarm center, and in another design enable automated updating or change of current silhouettes or shapes of contiguous areas to be detected. Said machine learning enables remote setting of the variable, which considerably simplifies installation work as well as maintenance and service, and enables remote updates in the event of circumstances or changes in the environment. A camcorder detector can be equipped with machine learning or for different types of installations. The video camera detector can thus be given different functions depending on needs or desires, for example an ability to switch between IR and visible light, or between maintaining a pure detector of objects to be able to vary its function between being a single detector and being a video camera for transmitting images. , image sequences or digital film. The aforementioned machine learning provides both the opportunity for automatic updates without the installer or service staff having to visit the property, and the opportunity for changed or increased capabilities for detection based on the customer's wishes and future product development. Said machine learning also provides the possibility to vary the camera function as needed, such as to switch between IR detection and detection of light intensity, or to switch between detection and video display in grayscale or color, which enables switching between visible video image and detection.

A video camera detector can be installed as an incident detector for detecting persons who end up in different types of risk situations, such as at risk of drowning, the said incident detector detects and follows people as long as they are in one water, such as a pool, one or an outdoor pool. Said incident detector can in one design detect when someone ends up below the water surface without coming back up into the air, and in another design detect if a person has their face below the water surface for a longer period of time, and in another design if a person appears lifeless with the face down. Different designs of different risk detection when people or other objects move can be implemented in a video camera detector which can then effectively detect different accident risks with the possibility of alarms in connection with detected incidents.

The present invention means that fully automated video detection and / or video verification can take place completely without anyone watching the video and if it is not done, there will be video detection and video verification even on properties with a high degree of confidentiality, for example an alarm operator can be installed as in the defense. the judiciary or social services, as well as in private homes where residents value high security against intrusion while maintaining their privacy without transparency or risk of unwanted surveillance.

The present invention may also include that selected objects are selected or blocked from being detected, which may reduce the risk of error detection. detection, the security against error detection can be increased, for example a large dog can in some cases risk being detected as a human, but if a large dog can be given a form to opt out, the security against error detection can be reduced.

Said software and other functions mentioned above may in one design be implemented in a camcorder detector and in another design be implemented in a receiving unit for processing images received from one or more camcorder detectors or from digital cameras of various kinds. The board receiving unit can be a central unit, an alarm unit, a buffer unit, alarm centers or any other unit for receiving and handling digital images for detecting objects.

Said receiving unit may be equipped with software routines for handling one or more camcorder detectors which have or have not been equipped with the possibility of external control with or without feedback from the connected receiving unit or digital camera of any kind. Said control with or without feedback may consist of changed image resolution, changed frame rate, changed image format, reduction or increase of number of rows or columns, removal or reintroduction of rows, columns or groupings of pixels, software update for changed functions of said camcorder detector or digital camera of any kind. or other types of external control or regulation of connected detectors.

Claims (8)

1. Device for detecting predefined objects moving in a monitored environment by automated detection of said object, characterized in that said device comprises a digital image acquisition unit (11) for capturing images, a computer unit (12) for handling said images with detection of changes between images in an image series and assessment if said change corresponds to said object, and a communication unit (13) for communication with the outside world. Said detection is based on changes to the content of a stream of digital images and said changes are examined through comparisons between said changes and one or more predetermined shapes corresponding to said predefined objects. Said image recording unit, said computer unit and said communication unit can in different configurations be part of each other, be part of a single physical unit, or be divided into more than one physical unit. 2. Device for automated detection in claim 1, characterized in that said stream of digital images has a lower frame rate than what a digital video camera delivers, down to one or down to one or a few frames per second (FPS), and that said lower frame rate can be created by extracting said stream of digital images with a lower frame rate into a digital image stream with a higher frame rate. 3. Device for automated detection according to known features in that said stream of digital images can be varied from less than one image per second or from one or a few images per second, up to a larger number of images per second. 4. Device for automated detection in accordance with requirement 4, characterized by the fact that said variation of image frequency can be controlled or regulated from an externally connected unit. 5. Device for automated detection e n I i gt kra v 1 kn e t eck n a dav that the resolution in said digital images can be varied between a higher number of pixels per image to a lower number of pixels per image. 6. Device for automated detection according to requirement 4 characterized by the fact that said resolution can be controlled or regulated from an externally connected device. Device for automated detection according to requirement 1 characterized by the possibility of variation of the number of pixels to be examined and rows of pixels, columns of pixels or blocks of pixels can be selected from being examined, and said variation can be controlled or regulated from an externally connected device. 7. Method for automated detection of changes in the content of a stream of digital images according to claim 1, characterized in that a reference image based on several previous images is calculated and stored as a reference image, and that the current image is compared against said reference image and the difference is created as a difference image. Said difference image is then compared with one or more predetermined shapes for detection and if there is sufficient agreement this is considered as having found a searched object and a signal for positive detection of a searched object is created. Method for handling predetermined shapes to be used for investigation of possible correspondence with detected changes in a stream of digital images according to claim 1, characterized in that said handling means that said forms are changed or updated from an externally connected unit. 8. Device for automated detection as claimed in claim 1, characterized in that said handling of said images with detection images in receiving unit for detection of said object by processing of changes between an image series, are implemented in a received images and image series.