Identity-hiding motion detection and depiction device

Abstract

An identity-hiding motion detection and delineation device for privacy-preserving surveillance or monitoring, concealing the identities of detected moving objects and their observed locations, and denying access to the original video frames. The device includes a video camera that collects video frame images of an area and a volatile memory that stores a video buffer, where the video camera stores a stream of video frames in the buffer. For each pair of consecutive video frames in the video buffer, a processor calculates a simple difference frame of the pair, erases the first video frame of the pair from the volatile memory, and outputs the difference frame as a delineated motion video. The device may include an analysis module for detecting specific events, along with an alert module that issues an alert to an external device. The difference frames and alerts may pass through a one-way "waterfall" link within the device to prevent access to the original video frames.

Description

Technical Field

[0004] , , , ,

[0003]

[0001] The present invention relates to the field of video analysis for observation and surveillance, and more particularly to an apparatus for detecting and depicting motion captured in video image frames while concealing the identification of subjects within the images.

Background Art

[0002] "Smart" motion detection devices employing video cameras are used in public places and in private homes to warn of intrusion, the presence of unauthorized persons, and danger. The detection devices include motion analysis and object classification using morphology and other well-known techniques.

[0003] U.S. Patent No. 5,969,755 discloses a method for providing automatic content-based video indexing from object motion. Moving objects in video from a surveillance camera are detected within a video sequence using a motion segmentation method. The objects are tracked through the segmented data. A symbolic representation of the video is generated in the form of annotated graphics that describe the objects and their motion. A motion analysis device analyzes the results of object tracking and annotates the motion of the graph with an index that describes several events. The graph is then indexed using a rule-based classification method to identify events of interest such as appearance / disappearance, attachment / removal, entry / exit, and motion / stillness of the objects. Video clips identified by spatio-temporal, event, and object-based queries are played back to display the desired video.

[0004] U.S. Patent No. 6,049,363 discloses object detection for scene change analysis, performed by a statistical test applied to data extracted from two images taken from the same scene from the same viewpoint. It is assumed that a single change region is given corresponding to an object that is present in one image but not in the other. For TV data, the test consists of measuring the matching of edge pixels in each image that have a boundary for the change region. For IR data, the test consists of measuring the change in pixel intensity within the change region in each image. [Overview of the project] [Problems that the invention aims to solve]

[0005] So-called "stupid" motion detection devices, such as those employing passive infrared (PIR) sensors, do not clearly identify the detected subject (people and objects). Therefore, they are permitted for use almost anywhere. However, existing "smart" motion detection and warning devices are camera-based, and because they infringe on the privacy of those being filmed, they present legal and regulatory conflicts in many countries.

[0006] Regulations (such as those imposed by the Helsinki Human Rights Commission) are not the only obstacles to the use of such smart devices. Naturally, ordinary people are reluctant to install such devices in their homes, as they don't want to find video clips of them secretly captured and distributed across the web and social networks.

[0007] Smart motion detection devices focus on analyzing motion within video frames, detecting it, and classifying detected objects. They automatically issue fast alerts (either by human or machine). However, existing smart detection devices are not plagued by privacy issues or restrictions, such as the prohibition and / or undesirability of disclosing locations shown in photographs.

[0008] An effective alert is one with zero false alarms and zero missed actual alarms. The best possible alert would probably be one that transmits a photo of the event triggering the alert in real time, so it seems there is no way to have the optimal warning device without infringing on the privacy of the subject and location shown in the photo (a detailed photo that most people would not be happy to share). [Means for solving the problem]

[0009] This invention relates to an anonymous motion detection and recording device. The device prevents any leakage of detailed images or videos—therefore, a privacy violation—by discarding image data acquired by the device during data processing while saving and transmitting only the motion recording.

[0010] In exemplary embodiments, the anonymous motion detection and rendering device does not permit the passage of any images, such as video information. Images are used for motion detection processing and are then discarded by deletion or erasure from the device's memory. Images cannot be accessed for display or transmission. Only processing data of the movement of moving objects monitored by the device's field of view can be stored and shared by the device.

[0011] The present invention therefore provides an identification-concealing motion detection and depiction device for privacy-protecting surveillance and / or supervision by concealing the identification of detected moving subjects and their observed positions and denying access to the original video frames, the device, a. Video camera; b. A volatile memory on which a video buffer is stored; the video camera is configured to store a stream of video frames within the video buffer; c. For each pair of consecutive video frames in the video buffer, i. Calculate the difference frame of a pair; ii. Clearing the first video frame of a pair from volatile memory; iii. Output the differential frames as a motion video. A processor configured to perform Includes.

[0012] The present invention further provides the aforementioned apparatus, wherein the processor is further configured to smooth the edges of the depicted motion within the differential frame, to present a symbolic graphic image of the moving subject, or a combination thereof.

[0013] The present invention a. Real-time alerts for detected moving objects or multiple objects; b. Difference frames; and c. Symbolic graphic illustration of a moving object Further providing any one of the aforementioned devices, further including a wireless communication module (such as WiFi or cellular 3G / 4G / 5G) configured to transmit any combination of the above.

[0014] The present invention further provides any one of the aforementioned devices configured for a setup that allows the device to be aimed at a desired field of view without revealing an actual image of the position during setup.

[0015] The present invention further provides any one of the aforementioned devices, further configured to dynamically change the frame rate of the analyzed video frames by constantly comparing motion estimations of the same video sequence applied simultaneously to pairs of frames over short and long time intervals, and adjusting the frame rate accordingly if similarly fast or slow motion is detected.

[0016] The present invention further provides any one of the aforementioned devices further configured to apply pixel acceleration motion detection, wherein each pixel value is replaced by a suitable acceleration measurement, such that it is estimated by taking the second derivative of an interpolation curve obtained from the previous N frames.

[0017] The present invention further provides any one of the aforementioned devices, and the video camera is separated from the rest of the device and connected via any wired or wireless communication such as USB or MIPI.

[0018] The present invention further provides any one of the aforementioned devices, which further includes a video analysis module configured to detect events calculated from differential frames, video frames, or a combination thereof.

[0019] The present invention further provides any one of the aforementioned devices, and the events include the presence of an intruder, a fire alarm, face recognition, a fall, violent behavior, or any combination thereof.

[0020] The present invention further provides any one of the previous two devices, which is further configured to transmit an alert of an event to an external device.

[0021] The present invention further provides the previous device, and the communication path between the analysis module(s) and the alert module includes a one-way waterfall data link.

[0022] The present invention a. A microphone; b. An audio buffer configured to store audio signals collected by the microphone; c. An audio stamp database for storing audio stamps of event sounds; d. An audio analysis module configured to identify audio events stored in the audio buffer by comparison with audio stamps The present invention further provides any one of the previous four devices, which further includes an audio enhancement device including

[0023] The present invention further provides the previous device, and the results of the audio analysis module are correlated with the results of the video analysis module.

[0024] The present invention further provides an identification-concealing motion detection and depiction device for privacy-protecting surveillance and / or supervision by concealing the identification of detected moving subjects and their observed positions and denying access to the original video frames, the device, a. A video camera configured to collect video frame images of a region; b. Section 1, including the following: i. First video buffer: The video camera is configured to store a stream of video frames in the video buffer; ii. For each pair of consecutive video frames in the video buffer, a) Compute the difference frames of the pair; and b) Output the differential frames as a motion video. A processor configured to perform the following actions: c. Section 2, including the following: i. A second video buffer configured to store a stream of differential frames; ii. A second processor configured to transfer differential frames to a video encoder, the video encoder configured to encrypt the differential frames and output them to an external network; Includes, A one-way waterfall data link transports differential frames from the first section to the second section.

[0025] The present invention further provides the previous device, and the waterfall link is a. The transmitter in the first section and the receiver in the second section; b. One-way serial connection; c. Unidirectional optical fiber; and d. Analog video link Includes one or more of the following.

[0026] The present invention further provides any of the aforementioned devices having a waterfall link, wherein the first processor is further configured to erase a pair of first video frames from a first video buffer.

[0027] The present invention further provides any of the aforementioned devices equipped with a waterfall link, a. The first section further includes a first analysis module configured to detect events calculated from video frames; and / or b. The second section further includes a second analysis module configured to detect events calculated from the difference frame; and c. The second section further includes an alert module configured to send event alerts to an external device.

[0028] The present invention further provides the preceding apparatus, wherein the communication path between the analysis module(s) and the alert module includes one or more additional one-way waterfall data links.

[0029] The present invention further provides any of the aforementioned devices equipped with a waterfall link, wherein the first section receives software updates via a non-wireless connection.

[0030] The present invention provides an identification-concealing motion detection and depiction method for privacy-protecting surveillance and / or supervision by concealing the identification of detected moving subjects and their observed positions and denying access to the original video frames, the method a. The video camera collects video frame images within the area; b. Storing a stream of video frame images in a volatile memory video buffer; c. For each pair of consecutive video frames in the video buffer, i. Calculate the difference frame of a pair; ii. Clearing the first video frame of a pair from volatile memory; iii. Output the differential frames as a motion video. This includes the following steps.

[0031] The present invention provides the aforementioned method, further comprising one or more steps of smoothing the edges of depicted motion within a difference frame, presenting a symbolic graphic representation of a moving subject, or a combination thereof.

[0032] This invention provides a wireless communication module, a. Real-time alerts for detected moving objects or multiple objects; b. Difference frames; and c. Symbolic graphic illustration of a moving subject Further providing any one of the aforementioned methods, further including the step of sending any combination of .

[0033] The present invention further provides any one of the aforementioned methods, further comprising a setup for aiming the device into a desired field of view without revealing an actual image of the position during setup.

[0034] The present invention further provides any one of the above methods, further comprising the step of dynamically changing the frame rate of the analyzed video frames by constantly comparing motion estimations of the same video sequence applied simultaneously to pairs of frames over short and long time intervals, and adjusting the frame rate accordingly if similarly fast or slow motion is detected.

[0035] The present invention further provides one of the above methods, further comprising the step of applying pixel acceleration motion detection, wherein each pixel value is replaced by a suitable acceleration measurement thereof, such that it is estimated by taking the second derivative of an interpolation curve obtained from the previous N frames.

[0036] The present invention further provides any one of the above methods, further comprising the step of providing a video camera that is separated from the rest of the apparatus and connected via wired or wireless communication.

[0037] The present invention further provides any one of the above methods, further comprising the step of a video analysis module detecting an event calculated from a difference frame, a video frame, or a combination thereof.

[0038] The present invention further provides the previous method, wherein the events include the presence of an intruder, a fire alarm, facial recognition, a fall, an act of violence, or any combination thereof.

[0039] The present invention further provides one of the two preceding methods, further comprising the step of sending an event alert to an external device.

[0040] The present invention a. Microphone; b. An audio buffer configured to store the audio signal collected by the microphone; c. A sound stamp database for storing sound stamps of events; d. A speech analysis module configured to identify speech events stored in the speech buffer by comparing them with speech stamps. The present invention further provides one of the three preceding methods, further comprising the step of providing a voice enhancement device.

[0041] The present invention provides the prior method, further comprising the step of correlating the results of a voice analysis module with the results of a video analysis module. [Brief explanation of the drawing]

[0042] [Figure 1] This is a sharp depiction of the outline of a moving subject within the field of view of the device, generated according to several embodiments of the present invention. [Figure 2] This is a depiction in which segments of recognized moving subjects are replaced with symbolic graphic images. [Figure 3] This is a functional block diagram of an identification and concealed motion detection and rendering device according to several embodiments of the present invention. [Figure 4] This illustrates a process where the earlier of two video frames generating a difference frame is erased from RAM after the difference frame has been calculated. [Figure 5] The images show, respectively, a location without moving subjects and a non-disclosed view of an image generated according to several embodiments of the present invention. [Figure 6] The images show, respectively, a location without moving subjects and a non-disclosed view of an image generated according to several embodiments of the present invention. [Figure 7] This is a functional block diagram of an identification-confidential motion detection and drawing device with AI analysis, according to several embodiments of the present invention. [Figure 8] This is a functional block diagram of a motion detection and rendering device that uses AI analysis of specific events and features based solely on differential images, according to several embodiments of the present invention. [Figure 9] This shows a comparison between an image taken from a scene where the subject has fallen and an image taken at the time of capture that conceals the scene's identification and motion. [Figure 10] This document presents morphological signatures of dogs, women, and men that can be compared to Identification-Concealed Motion Descriptions, which can be used to determine the type of subject moving within the description. [Figure 11] This is a functional block diagram of an identification-confidential motion detection and rendering device according to several embodiments of the present invention, in which AI analysis of specific events and features is performed from complete visual data of video frames and sequences. [Figure 12] This is a functional block diagram of a voice enhancement 500 that can be used with an identification concealment motion detection and drawing device, according to some embodiments of the present invention. [Figure 13] This is a functional block diagram of an identification-confidential motion detection and drawing device 600 with a one-way "waterfall" data link, according to some embodiments of the present invention. [Modes for carrying out the invention]

[0043] In this disclosure, the term “motion portrayal” refers to providing an image that facilitates the detection of motion (by a person or machine). In exemplary embodiments, the provided image highlights the edges of a moving object.

[0044] The term "full visual video data" or simply "full visual data" refers to the unprocessed video frames as they were initially captured by a video camera or repetitive still camera.

[0045] "Visual data" may also refer to video frames that have been processed to extract only the outlines of moving subjects, as will be further described herein.

[0046] "Non-visual data" refers to data about motion within a video (whether fully visible or fully visible) extracted from video frames.

[0047] In exemplary embodiments, the present invention includes an identity-confidential motion detection and rendering device that does not store acquired visual and / or IR video data in any externally accessible memory. The acquired video data is temporarily stored in the device's random access memory (RAM) for processing and is erased from the RAM immediately after processing.

[0048] In one possible embodiment of the present invention, N video frames are stored in the RAM of the device, and in between, the Nth frame is compared with consecutive frames (e.g., the N+1th, N+2nd, or N+nth frame) using motion estimation and image comparison techniques to detect the edges of any present moving objects.

[0049] A video camera samples its field of view several times per second, typically at 24–60 frames per second (FPS). In some embodiments, the device calculates a simple difference between consecutive video frames. The frame showing such a difference is called a difference frame (or simply a difference). In such embodiments, once the device is stationary (not panning, tilting, or zooming), it calculates a simple difference between consecutive frames—whether skipping to consecutive frames N+1, N+2, or N+n—to create a sharp depiction of the contours of moving subjects within the device's field of view, as shown in Figure 1. The depiction of motion, in terms of contour thickness, relates to the distance the moving subject has traveled between frames in the pixels. The larger the time interval between the differed frames, the thicker the contour lines of the difference image of the moving subject become. Thickness is also affected by the speed of motion; the faster the motion, the thicker the contour of the moving object becomes.

[0050] While video cameras typically capture 24–60 FPS, for the present invention, a lower frame rate of 1–12 FPS is typically sufficient. However, large time intervals (e.g., above 100 ms) can generate difference images that reveal identifying details of fast-moving subjects. Enhancing the contrast of the difference images and / or reducing the brightness would be one way to discard such details to conceal the identification of moving objects. Another method might be to reduce the color depth to 8, 4, or even 2 colors. Additionally (or alternatively), one of the many edge detection techniques well known in the art could smooth the contours and thin the lines, and / or even vectorize the difference images to reduce data size / rate for optimized transmission.

[0051] As shown in Figure 2, another method of concealing identification is to replace the recognized moving object with a symbolic graphic illustration 50 of the moving subject.

[0052] In some exemplary embodiments, the anonymous motion detection and rendering device does not allow the passage of any images. Video information is used for motion detection processing and discarded, deleted / erased from the device's memory, and inaccessible for display or transmission. Only processing data of the movement of moving objects monitored by the device's field of view can be stored and shared by the device.

[0053] Herein, we refer to Figure 3, a functional block diagram of an identification-confidential motion detection and drawing device 100 according to several embodiments of the present invention. The device 100 includes a video camera 105, one or more volatile memory (RAM) 110, one or more processors 115 (collectively, the "processors"), a video encoder 130, communication means 135, and one or more storage media 140.

[0054] The video camera 105 is a hardware-mounted camera sensor necessary for acquiring video frames and placing them into a video buffer 120 in RAM 105, and is typically nothing more. The video camera 105 is positioned to capture images of a monitored or observed area. Typically, the video camera 105 acquires image frames at a frame rate of 1 to 12 FPS. The video camera 105 may be an integrated component of the apparatus 100 according to the present invention, but the present invention can also be implemented by connecting a commercially available video camera (or a still camera capable of acquiring video) to a separate apparatus designed according to the present invention.

[0055] Typically, the processor 115 and the video encoder 130, which encodes the difference stream 125 to produce the output stream 132, are parts of a different device. The video encoder 130 cannot access the original video frames. The only data the video encoder 130 can access is the data of the difference image from the difference stream 125. The difference image is sufficiently obscured to maintain the privacy of the location and objects within the field of view of the device 100. In an alternative embodiment, the processor 115 includes the video encoder 130 (for example, the function of encoding the difference stream 125 to produce the output stream 132 is performed by the processor). Referring to Figure 4, in such an embodiment, the processor 115 implements a process in which the earlier of the two video frames that generate the difference frame is erased from RAM 110 after the difference frame has been computed. This process denies the video encoder 130 access to the original video frames.

[0056] Once a difference image is generated, the original video frames are erased from the video buffer 120. Since the difference stream 125 is generated, the original video frame data no longer exists and is therefore inaccessible. The difference frames can be encoded (at their original or reduced resolution) and wrapped as a video stream, which can then be transmitted via wired or wireless connection for remote monitoring of the location.

[0057] The present invention enables simple and effective alerts whenever a certain level of motion is detected. In some embodiments, the processor 115 counts the number of pixels measuring different light intensities between consecutive video frames in the video buffer 120. In some embodiments, the processor reduces the difference frame to a 1-bit color depth and counts the number of white or black pixels in the difference frame.

[0058] Those skilled in the art, after learning the teachings of this disclosure, will be able to, in accordance with the present invention, specify different regions within the monitored field of view of the device and, according to defined specifications, alert for or ignore movement within the specified regions.

[0059] The outlines of moving objects are classified based on simple morphology, and pets, birds, humans, etc., are distinguished by any means known in the art.

[0060] Since the device according to the present invention cannot output images of the acquired video, it outputs only the difference between consecutive frames (or any other depiction of the outline of a moving object within the device's field of view). The present invention makes it possible to simply set the field of view of the device without revealing a location-identifiable image.

[0061] In some embodiments, the device is in setup mode, and therefore the acquired frames are synthetically shifted by a few pixels (horizontally, vertically, or both) as if the entire scene were moving. Such synthetic motion allows for the output of a difference image or difference video stream of the contours of objects within the device's field of view.

[0062] Figure 5 shows a location without a moving subject. A static device according to the present invention displays a blank image (black, gray, white, or any other color) of the device's field of view and does not display any image. The resulting lack of reference features presents a problem in setting up the device to monitor a desired field of view.

[0063] As a possible solution to this problem, according to several embodiments, the processor implements a non-disclosed view of the scene. This view is obtained by shifting consecutively acquired frames by a few pixels horizontally and a few pixels vertically from frame N to frame N+1, and the same shift is applied again to shift from N+2 to N+3, and so on (one frame in the original position from which the subsequent frames were shifted), and so on. The result is shown in Figure 6.

[0064] Herein, we refer to Figure 7, a functional block diagram of an identification-confidential motion detection and depiction device 200 equipped with a video analysis module 222, according to several embodiments of the present invention. The analysis module 222 may employ artificial intelligence (AI), as shown in the figure.

[0065] The video analysis module 222 is introduced between the video buffer 220 and the differential stream 225, maintaining the elements and functions of the system as described in relation to Figure 3. Frames acquired by the video camera 205 are placed in the video buffer 220. Before the frames are compared to create a privacy-preserving differential image, the video analysis module 222 analyzes the video frames and short video sequences using state-of-the-art methods for face recognition, fall detection, motion loss, and / or other hazardous conditions.

[0066] The video analysis module 222 stores the analysis results, which are non-visual data. Such non-visual data may include, for example, in the case of face recognition, only the 2D and / or 3D common ratios and relative angles of the facial features of the subject being analyzed (eyes, nose, nostrils, forehead, eyebrows, ears, chin, hairline, etc.).

[0067] Herein, refer to Figure 8, a functional block diagram of the identification concealed motion detection and depiction device 300, in which video analysis of specific events and features 322 is based solely on difference images.

[0068] Event detection from difference images by the video analysis module 222 can be implemented by any means well known in the art. Since the subjects are separated from the static scene in which they are located, difference images can reduce the effort required to analyze the data. For example, if the colors of the objects and the background are similar, the static background is not present in the difference image in such cases, so it may be even easier to analyze the motion based on edge detection of pure differences between frames (however, in other cases, difference image analysis may be more complex due to the lack of detailed visual data. For example, the detection of falls and human posture can be performed more accurately when complete visual data is available, as described herein, for example, in relation to Figure 11).

[0069] Employing the device 300 to detect when a subject (intruder) enters the scene is a simple task: when they enter the scene, any means known in the art can be employed to filter out noise from the difference image and sum the number of groups of adjacent pixels (blobs) that exceed a threshold number of pixels.

[0070] Frames are placed from the video camera 305 into the video buffer 320 of the device's volatile memory. The frames are compared, and standard or enhanced difference image frames are created in the difference stream 325. Immediately after their creation, the original video frames in the video buffer 320 are erased (and overwritten).

[0071] In the next step, the sequence of such difference frames is processed to analyze the motion vector of the pixel groups by comparing the amount of moving pixels contained within a group of adjacent pixels (blobs) with the motion of such groups of pixels in the previous frame (if any).

[0072] For example, tipping detection can be implemented using well-known techniques, such as calculating the acceleration of a difference pixel moving vertically, during which the ratio of its horizontal to vertical dimensions in the rectangle surrounding such pixel changes significantly. A sudden movement of a fairly large number of difference pixels, as shown in Figure 9, combined with a change in the H:V ratio of the surrounding rectangle from V>H to V≦H, is highly likely to indicate a tipping.

[0073] According to the present invention, an alert can be verified by morphological comparison of signatures of moving objects in the scene (men, women, children, pets, or any predefined objects), as illustrated in Figure 10. Techniques for performing the comparison and matching such morphological stamps can be implemented by any means known in the art.

[0074] Referring here to Figure 11, a functional block diagram of an identification concealed motion detection and depiction device 400 is shown, in which, according to some embodiments of the present invention, analysis of specific events and features 422 is performed from the complete visual data of video frames and sequences while they are stored in a video buffer 420, before creating differential frames and discarding the complete visual data.

[0075] The complete video analysis module 422 may be a standalone analysis component of the device 400, or it may be used as a decision support component, used in combination with differential frame analysis, as described herein (e.g., the differential frame analysis module 322 in Figure 8). Verification of reliable identification of trigger events by correlating the analysis of the complete video frame with the analysis of the differential frame yields more accurate results and minimizes the rate of false positive and false negative identification.

[0076] Here, refer to Figure 12, a functional block diagram of the audio enhancement device 500, which can be used with the identification concealment motion detection and depiction device.

[0077] Similar to video, the present invention conceals the identification of the location and person being monitored by preventing the streaming of audio from the device.

[0078] Nevertheless, the device 500 is designed to identify specific events by comparing outlying audio signals with stored audio stamps 565 stored within the device 500. Such stored audio stamps 565 may include, among other stamps, various sounds of falls, trigger sounds and / or words, which may also be added by recording user / multiple users.

[0079] Once an audio trigger event is detected, it can be correlated with video analysis results from the same time period, if there are any doubts, to minimize false negative and false positive alerts.

[0080] Herein, Figure 13 shows a functional block diagram of an identification-confidential motion detection and drawing device 600 with a one-way "waterfall" data link 670, according to several embodiments of the present invention.

[0081] The apparatus 600 is divided into a first section 602 and a second section 604. The first section 602 performs a first (initial) phase of acquiring video frames. A processor 615 in the first section calculates differential frames. The first section may include an analysis module 622 that performs analysis with respect to the complete visual video frame. The differential frames are supplied to the frame buffer 620' of the second section 604 as source video frames for the second section 604. The differential frames are supplied through a one-way "waterfall" link 670. In the illustrated embodiment, the waterfall link 670 is implemented by a single transmitter 672 in the first section 602 and a single receiver 674 in the second section 604. In alternative embodiments, the waterfall link may be implemented by a one-way serial connection, through a one-way optical fiber, as analog video (converted from digital to analog and transmitted through the cable, and then, if necessary, digitized, encoded and broadcast) and / or similar one-way means.

[0082] The video encoder 630 of the second section 604 encodes the difference image as a video stream 632 and broadcasts the stream 632 as needed. The second section 604 further issues alerts according to a full visual video analysis (by the analysis module 622 of the first section 602) and / or a difference image analysis (by the analysis module 622' of the second section 604).

[0083] The differential frame is the sole visual output of the first section 602 to the second section 604. Furthermore, since the waterfall link 670 is unidirectional and the external network 635 is connected only to the second section 604, the first section 602 cannot receive external requests for complete visual data from the network 635 (only the downstream portion of the device 600 from the waterfall link 670 can be accessed from the outside). Thus, a hacker cannot access and steal the complete visual data, and cyberprivacy is protected therein.

[0084] In a preferred embodiment, both the first section 602 and the second section 604 each have independent memory and independent processors 615, 615', each including video buffers 620, 620'.

[0085] In some embodiments, the video encoder 630 encrypts the output video of the second section 604, thereby requiring decryption at the client end.

[0086] In some embodiments, the analysis module 622 of the first section 602 is connected to the first analysis module 622 to receive software updates via a non-wireless connection, such as an SD card or USB, thereby eliminating the need to connect to a network. Preferably, the analysis module 622 of the first section uses encrypted files.

[0087] In the apparatus 600 of Figure 13, the waterfall link 670 is located in the connection that carries differential frames. Alternatively, or additionally, a one-way waterfall link may be located in the connection that transmits the video stream 632 and / or alerts 645 to the external network 635. For example, the video encoder 630 may have only a WiFi transmitter to the external network 635 and no receiver.

[0088] The alert module 645 may receive embedded video signals, such as colored pixels, such as an 8x8 pixel macroblock in one of 16 or 24 colors, where each color is associated with a specific event, via a waterfall link. Only the alert module of the device is likely accessible from external devices 650 (or their network). The upstream portion of the device from the waterfall that carries the alerts is isolated from external access.

[0089] The waterfall link that carries the alerts can be implemented, for example, by a serial transmission cable.

[0090] [Table 1]

[0091] It should be understood that features presented with different reference numbers but sharing a common name are not necessarily identical.

Claims

1. 1. An identity-hiding motion detection and depiction device for surveillance and / or supervision that preserves privacy by hiding the identities of detected moving objects and their observed positions and denying access to original video frames, the device comprising: a. a video camera configured to collect video frame images of an area, said video camera being an integral component of said device or a separate device and connected to said device or said separate device via wired or wireless communication; b. a volatile memory on which a video buffer is stored, the video camera being configured to store the stream of video frames in the video buffer; c. a processor, for each pair of consecutive video frames in said video buffer: i. calculating said pair of difference frames; ii. erasing a first video frame of the pair from the volatile memory; iii. outputting the difference frames as a rendered motion video; A processor configured to: An apparatus comprising:

2. a. a real-time alert of one or more detected moving objects; b. The difference frame; c. A symbolic graphic depiction of a moving object; 13. The apparatus of claim 1, further comprising a wireless communication module (such as WiFi or cellular 3G / 4G / 5G) configured to transmit any combination of.

3. The apparatus of claim 1, wherein the apparatus is configured for setup to enable aiming of the apparatus at a desired field of view without revealing an actual image of the location during setup.

4. The apparatus of claim 1, further configured to dynamically vary the frame rate of the analyzed video frames by constantly comparing motion estimates of the same video sequence applied simultaneously to pairs of frames spanning short and long time intervals, and adjusting the frame rate accordingly if similarly faster or slower motion is detected.

5. The apparatus of claim 1, further configured to apply pixel acceleration motion detection, wherein each pixel value is replaced by its appropriate acceleration measurement, as estimated by taking the second derivative of an interpolated curve obtained from the previous N frames.

6. a video analytics module configured to detect events computed from the difference frames, the video frames, or a combination thereof, the events including presence of an intruder, a fire alarm, facial recognition, a fall, an act of violence, or any combination thereof; and (b) an alert module configured to send an alert of the event to an external device.

7. The apparatus of claim 6 , wherein a communication path between the video analytics module and the alert module comprises a unidirectional waterfall data link.

8. The apparatus further comprises a voice enhancement device, the voice enhancement device comprising: a. a microphone; b. an audio buffer configured to store audio signals collected by the microphone; c. a sound stamp database for storing sound stamps of sound events; d. a voice analysis module configured to identify voice events stored in the voice buffer by comparison with the voice stamps; Equipped with The apparatus of claim 6 , wherein results of the audio analysis module are correlated with results of the video analysis module.

9. 1. An identity-hiding motion detection and depiction device for surveillance and / or supervision that preserves privacy by hiding the identities of detected moving objects and their observed positions and denying access to original video frames, the device comprising: a. a video camera configured to collect video frame images of an area; b. a first section, i. a first video buffer, wherein the video camera is configured to store the stream of video frames in the first video buffer; ii. a first processor, for each pair of consecutive video frames in the video buffer: a) calculating said pair of difference frames; b) outputting the difference frames as rendered motion video; A first processor configured to: a first section including: c. A second section, i. a second video buffer configured to store the stream of difference frames; ii. A video encoder; iii. A second processor configured to forward the difference frame to the video encoder, the video encoder configured to encrypt and output the difference frame to an external network; a second section including: Equipped with An apparatus, wherein a unidirectional waterfall data link carries the difference frames from the first section to the second section.

10. The waterfall link is a. a transmitter in the first section and a receiver in the second section; b. A unidirectional serial connection; c. A unidirectional optical fiber; d. an analog video link; The apparatus of claim 9 , comprising one or more of:

11. the first processor is further configured to erase a first video frame of the pair from the first video buffer; the first section further comprises a first analysis module configured to detect events computed from the video frames; and / or b. the second section further comprises a second analysis module configured to detect events calculated from the difference frame; and 10. The device of claim 9, wherein the second section further comprises an alert module configured to send an alert of the event to an external device.

12. The apparatus of claim 11 , wherein a communication path between at least one of the first and second analysis modules and the alert module includes one or more additional unidirectional waterfall data links.

13. 1. An identity-hiding motion detection and delineation method for surveillance and / or supervision that preserves privacy by concealing the identities of detected moving objects and their observed positions and denying access to original video frames, the method comprising: a. obtaining a device according to claim 1; b. the video camera collecting video frame images within the area; c. storing said stream of video frame images in a video buffer of said volatile memory; d. for each pair of consecutive video frames in said video buffer, i. calculating said pair of difference frames; ii. erasing a first video frame of the pair from the volatile memory; iii. outputting the difference frames as a rendered motion video; Including, The method further includes one or more steps of smoothing edges of the depicted motion in the difference frame, presenting a symbolic graphic depiction of the moving object, or a combination thereof; The method includes: a. a real-time alert of one or more detected moving objects; b. The difference frame; c. A symbolic graphic depiction of a moving object; d. Aiming the device to a desired field of view without revealing an actual image of the location during the setup; transmitting any combination of.

14. - dynamically varying the frame rate of the analyzed video frames by constantly comparing motion estimations of the same video sequence applied simultaneously to pairs of frames spanning short and long time intervals, and adjusting the frame rate accordingly if similarly faster or slower motion is detected; applying pixel acceleration motion estimation, where each pixel value is replaced by its appropriate acceleration measurement, as estimated by taking the second derivative of an interpolation curve obtained from the previous N frames; providing that the video camera is separate from the rest of the device and connected via wired or wireless communication; The method of claim 13 further comprising: a video analytics module detects events calculated from the difference frames, the video frames, or a combination thereof, the events including presence of an intruder, a fire alarm, facial recognition, a fall, an act of violence, or any combination thereof; b. sending an alert of said event to an external device; The method of claim 13 further comprising:

16. The method of claim 15 further comprising providing a one-way waterfall data link along a communication path between the analysis module and the alert module.

17. The method further comprising the step of providing a voice enhancement device, the voice enhancement device comprising: a. a microphone; b. an audio buffer configured to store audio signals collected by the microphone; c. a sound stamp database for storing sound stamps of sound events; d. a voice analysis module configured to identify voice events stored in the voice buffer by comparison with the voice stamps; Equipped with The method of claim 15 , further comprising correlating results of the audio analysis module with results of the video analysis module.

Images