Event video sequence

Abstract

Event video sequences are disclosed. A method of providing an event initiated video sequence, the video sequence comprising a pre-event video sequence and a post-event video sequence. Pre-event image frames are stored in an unprocessed format until such time as an event occurs. By deferring image processing of the pre-event video sequence, power savings are achieved.

Description

Technical Field

[0001] This invention relates to capturing and processing video sequences initiated by events. Background Technology

[0002] Video cameras, such as surveillance cameras or monitoring cameras, are currently used in many different application areas, both indoors and outdoors. Larger image sensors and other technological advancements have greatly improved the image quality of video from such cameras, but at the same time, the size of the video stream, or the bit rate expressed differently, has also increased. Since both storage capacity and network bandwidth are limited resources, the usual practice is not to store or stream all video, but only to store or stream video that is initiated or triggered by an event.

[0003] The term “event” is used here in its broadest sense; therefore, an event can be anything that might make the need for video documentation necessary, and the term can cover anything from a burglar alarm sounding when a door is forcibly opened to a visit by a care professional in a patient’s home.

[0004] It is typically of interest not only to make video available from the point in time when the event begins, but also to record video showing the monitored area or scene during the period exactly before the event begins. This is usually achieved using a so-called pre-event video buffer, in which video of the most recent time interval is recorded consecutively. When the event begins, the video recorded in the pre-event buffer is retrieved and included in a video clip describing the event. The applicant of this application, US8676027, discloses a method related to such a pre-event buffer in a camera.

[0005] Surveillance cameras can be installed in various locations, typically using one cable connection for power and another for network access, or, in the case of Power over Ethernet (PoE), both via a single connection. However, wearable cameras (often referred to as body-worn cameras (BWC)) have recently gained popularity. For example, police or security personnel use wearable cameras to capture video and other data during patrols and incidents. These cameras are typically battery-powered. This means that the available power of wearable cameras is much more limited than that of cameras mounted in fixed locations. Therefore, the means and measures that are very reliable when used with cable-connected cameras (such as those mounted on buildings) can be problematic with BWCs, as more attention needs to be paid to power consumption. One challenge is that in the area where an event is initiated and recording takes place, using an event pre-buffer requires the camera to continuously record video. This can quickly drain the battery, and therefore improvements in this area are of great interest. Summary of the Invention

[0006] One object of the present invention is to promote power saving in cameras associated with event-triggered video.

[0007] Based on the first aspect, a method for recording event video sequences is proposed, wherein the event video sequences include:

[0008] The pre-event video sequence contains video showing the monitored scene during a predetermined time period prior to event detection, and

[0009] The post-event video sequence contains videos showing the monitored scene starting from the time the event was detected.

[0010] The method includes the following steps:

[0011] Capture image frames continuously.

[0012] Newly captured image frames are continuously stored in the pre-event buffer by adding them to the set of pre-event image frames corresponding to a predetermined time length.

[0013] Receive the first signal indicating that an event has been detected.

[0014] After receiving the first signal

[0015] Aborts storing newly captured image frames in the pre-event buffer.

[0016] The newly captured image frames are fed to the image processing unit.

[0017] Image processing of the newly captured image frame is performed in the image processing unit, and

[0018] The processed image frames are stored in memory as a post-event video sequence.

[0019] Receive a second signal indicating that the event has ended.

[0020] After receiving the second signal,

[0021] Stop feeding newly captured image frames to the image processing unit.

[0022] The set of image frames is fed from the pre-event buffer to the image processing unit.

[0023] Image processing of a set of image frames is performed in the image processing unit, and

[0024] The processed set of image frames is stored in memory as a pre-event video sequence.

[0025] By storing a set of image frames in a pre-event buffer instead of processing them first in the image processing unit, power consumption by the camera can be significantly reduced during periods when no event is occurring, while still allowing the generation of pre-event video sequences when needed. This provides valuable insights into the actions occurring in the time period prior to the receipt of the first signal. By prioritizing the processing of the post-event video sequence over the processing of the pre-event image frame set, the post-event portion of the event video sequence can be streamed from the camera for real-time or near-real-time viewing by the user. The pre-event video sequence showing the scene during a finite time period prior to the first signal is processed immediately after the second signal is received and the event ends, and can be viewed shortly thereafter.

[0026] In battery-powered cameras, deferring image processing of the set of images forming the pre-event video sequence until the actual need prevents unnecessary battery drain. Therefore, the battery lasts longer, and the camera can be used for extended periods without recharging. This method will also be available in cameras where total power consumption needs to be kept within specific limits for other reasons, such as when the camera is powered via Power over Ethernet (PoE) and needs to be kept within predetermined power levels. A potential drawback is the need for a larger pre-event buffer, as unprocessed (raw) format image frames take up more space than processed (especially encoded) format image frames; however, this approach would be an attractive option when power consumption reduction is critical.

[0027] Another effect of delaying processing and subsequently storing the image in memory is that there will be fewer read / write operations performed on the memory, which in turn will increase the memory's lifespan.

[0028] Furthermore, deferred processing means that processing of image frames from the pre-event buffer may not need to be performed at full frame rate, allowing processing of each image frame to continue for a longer period. This makes it possible to utilize more advanced image processing and encoding algorithms without increasing processing power. Allowing processing to occur at a rate lower than full frame rate also means that processing can occur at a lower clock frequency, which reduces the total power required.

[0029] The method may also include the step of continuously discarding the oldest image frames from the pre-event buffer after storing newly captured image frames; or in other words, a circular buffer (first-in, first-out (FIFO)) memory may be used for the pre-event buffer. This provides an efficient option for storing an unprocessed set of image frames that constitute a potential pre-event video sequence.

[0030] The method may further include feeding a subset of image data from the newly captured image frame to the image processing unit before receiving the first signal. The subset of image data can be used to stream to a remote user to provide an overall overview of the scene (e.g., during a person carrying a camera's entire visit to the scene). The subset of image data may additionally or alternatively be stored in memory for later use.

[0031] A subset of image data can be generated by spatially subsampling the newly captured image frames, such that image frames with a reduced resolution compared to the newly captured frames are fed to the image processing unit. Alternatively, a subset of image data can be generated by temporally subsampling the newly captured image frames, such that a stream of image frames with a reduced frame rate compared to the frame rate of the newly captured frames is fed to the image processing unit. Both alternatives provide a means of creating a subset of image data with only a very limited number that can be processed in the image processing unit without incurring any significant power consumption, while still allowing for an overall overview of the scene.

[0032] The step of storing newly captured image frames in the pre-event buffer may include compressing the image frames. This achieves more efficient storage of the set of unprocessed image frames, allowing a smaller pre-event buffer to be used for the same number of image frames stored therein. The compression used in this step will be a power-efficient compression (such as run-length encoding or JPG encoding), which will not significantly increase the camera's power consumption while significantly reducing the size of the image frames.

[0033] The method may further include, after feeding a set of image frames from the pre-event buffer to the image processing unit, reverting to continuously storing newly captured image frames in the pre-event buffer. In this way, a new set of image frames will be available, which in turn can form the basis for a new pre-event video sequence.

[0034] The steps of performing image processing may include one or more of the following: defect pixel correction, white balance, demosaic, matrixing, gamma correction, sharpening, noise filtering, scaling, and encoding.

[0035] Delaying image encoding until the event occurs, in turn, means that both the pre-event and post-event video sequences can be started with an intra-coded frame (I-frame), which in turn makes decoding and viewing the event video sequence much easier. In existing solutions where the pre-event video sequence is encoded sequentially, there is little chance of controlling the encoding to ensure that an I-frame has been encoded at the start of the event video sequence, and when viewing prior art event video sequences, decoding and viewing the event video requires returning to the beginning of the encoded group of pictures (GOP), which can sometimes take up to a minute. In this method, decoding and viewing will be able to begin precisely at the start of the event video sequence because of the fact that encoding of the pre-event video sequence can begin with an I-frame.

[0036] The first signal can be received from an image analysis unit, which is configured to detect events in the monitored scene based on captured image frames. Examples of processing used by such an image analysis unit include video motion detection, object recognition, foreground-background segmentation, and other methods for determining certain types of events of interest occurring in a scene based on video images.

[0037] Alternatively, the first signal can be received from an external event detection unit, which is configured to detect events in the monitored scene. The term "external" should be interpreted as outside the camera, operating independently of the camera, or separated from the camera (e.g., physically or logically separated from the camera). Such an external event detection unit may include sensors for detecting that a weapon has been pulled from its holster or motion sensors for sensing, for example, that a person carrying the camera has fallen, been pushed, or started running. The external event detection unit may also include various types of positioning sensors (such as those for receiving GPS signals) to indicate that the camera has entered an area where video recording is required. Other examples of events that can be detected by external event detection are changes in GPS connectivity or connectivity to communication networks (such as connections to Wi-Fi or cellular networks).

[0038] The external event detection unit may also include sensors (such as PIR sensors, radar, or lidar) capable of sensing motion or objects in the scene. In some cases, the external event detection unit may also be another camera. This other camera may be of the same type as the camera itself, or it may have a completely different configuration. As an example, if the camera employing the method proposed herein is a body-wearable camera, the other camera could be another similar body-wearable camera, or it could be a surveillance camera mounted to monitor the scene (such as a thermal imaging camera or a visual camera). This camera may be able to send a first signal to any nearby body-wearable camera.

[0039] As an alternative, audio or sound from the scene can be analyzed to find indications that an event is taking place that needs to be recorded on video (such as gunshots or loud screams).

[0040] The second signal can be generated as a response to no longer receiving the first signal. This could be, for example, when a first signal indicating that a person wearing a camera is running is received from a motion sensor, and the sensor indicates that the person has stopped running.

[0041] Alternatively, the second signal can be generated in response to a timer reaching a predetermined value, with the timer started upon receiving the first signal. This can be useful where it can be safely assumed that the duration of any event to be recorded on the video will be shorter than the predetermined value. The second signal can be generated by an external event detection unit based on some kind of data from any sensor used to generate the first signal, or the absence of such data.

[0042] The method may further include, upon receiving a first signal indicating that an event has been detected, streaming processed image frames of a post-event video sequence to a receiver. In this way, a third party (such as a security operator) can be immediately notified that an event is occurring in the scene and can take appropriate action (such as sending backups to personnel present in the scene). The video can be streamed to a remote server, for example, via a mobile phone network or other type of wireless network connection. The security operator can view the streaming images via a video management system connected to the remote server.

[0043] According to a second aspect of the invention, a surveillance camera configured to record video sequences of events is provided.

[0044] The event video sequence includes:

[0045] The pre-event video sequence contains video showing the monitored scene during a predetermined time period prior to event detection, and

[0046] The post-event video sequence contains videos showing the monitored scene starting from the time the event was detected.

[0047] The cameras include:

[0048] The image capture unit is configured to continuously capture image frames.

[0049] The pre-event buffer is configured to continuously store newly captured image frames by adding them to a set of pre-event image frames corresponding to a predetermined time length.

[0050] The event unit is configured to receive a first signal indicating that an event has been detected and a second signal indicating that the event has ended.

[0051] The image processing unit is configured to process image frames.

[0052] The memory is configured to store processed image frames.

[0053] Among them, after receiving the first signal,

[0054] The camera is configured to stop storing newly captured image frames in the pre-event buffer and feed the newly captured image frames to the image processing unit.

[0055] The image processing unit is configured to perform image processing on newly captured image frames and store the processed images in memory as a post-event video sequence.

[0056] Among them, after receiving the second signal,

[0057] The camera is configured to stop feeding newly captured image frames to the image processing unit.

[0058] The camera is configured to feed a set of image frames from the pre-event buffer to the image processing unit.

[0059] The image processing unit is configured to perform image processing on a set of image frames.

[0060] The camera is configured to store a collection of processed image frames in memory as a pre-event video sequence;

[0061] Based on the aforementioned characteristics of the method in the first aspect, this also applies to the second aspect when applicable. To avoid unnecessary repetition, refer to the preceding text.

[0062] The further scope of the invention will become apparent from the detailed description given below. However, it should be understood that the detailed description and specific examples are given only by way of illustration when indicating preferred embodiments of the invention, as various variations and modifications within the scope of the invention will become apparent to those skilled in the art through this detailed description.

[0063] Therefore, it should be understood that the invention is not limited to the specific component parts of the described apparatus or the operation of the described method, as such apparatus and methods can vary. It should also be understood that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting. It must be noted that, as used in the specification and appended claims, “a,” “the,” and “the” are intended to indicate the presence of one or more elements, unless the context clearly specifies otherwise. Thus, for example, references to “a unit” or “the unit” can include several means, etc. Furthermore, the words “comprising,” “including,” “containing,” and similar wording do not exclude other elements or steps. Attached Figure Description

[0064] The invention will now be described more fully below with reference to the accompanying drawings, in which presently preferred embodiments of the invention are shown. However, the invention may be embodied in many different forms and should not be construed as limited to the embodiments set forth herein; rather, these embodiments are provided for thoroughness and completeness and to fully convey the scope of the invention to those skilled in the art.

[0065] Figure 1 A camera is shown.

[0066] Figure 2 Describing the use of Figure 1 The situation with the cameras.

[0067] Figure 3 This is a flowchart illustrating a method according to an embodiment of the invention. Detailed Implementation

[0068] Figure 1 A camera 100, such as a wearable or body-worn camera (BWC), is depicted. The camera 100 has optics 102 and an image capture unit 104 with an image sensor, as is generally known.

[0069] Back Figure 1 The camera 100 also includes an image processing unit 106 and a memory 108 for storing processed images. The memory 108 may be provided in the form of an SD card or flash memory. In the context of this application, the term "image processing" can encompass a variety of actions, some examples of which include defect pixel correction, white balance, demosaic, matrixing, gamma correction, sharpening, noise filtering, scaling, and encoding.

[0070] The camera also includes an event pre-buffer 112, which is configured to store a collection of event pre-image frames received from the image capture unit 104. The event pre-buffer may be provided in the form of a memory suitable for temporary storage, such as RAM, DRAM, or SRAM memory.

[0071] Typically, the pre-event buffer 112 is configured as a circular buffer or a first-in-first-out (FIFO) buffer, where a newly captured image frame fed from the image capture unit 104 to the pre-event buffer 112 replaces the oldest image frame already stored in the pre-event buffer 112. However, other configurations of the pre-event buffer 112 are also possible. Further improvements can be made by selecting which images to discard from the pre-event buffer. As an example, in the case of capturing HDR / WDR video, it would be possible to decide whether to discard all long exposure images or all short exposure images. If a longer period with a lower frame rate is considered useful when generating the pre-event video sequence, it would also be possible to discard each second image or each third image, etc.

[0072] The pre-event buffer 112 is positioned upstream of the image processing unit 106. Therefore, the set of pre-event image frames stored in the pre-event buffer 112 has not yet been processed by the image processing unit 106, meaning they will typically be larger than they would be after processing by the image processing unit 106, as is the case in prior art solutions for generating pre-event video sequences. However, as the inventors have realized, where power saving is critical, it is preferable to add more storage space to the pre-event buffer 112 to enable the storage of a set of raw or near-raw image frames from the image capture unit 104, rather than wasting power on unnecessary processing of image data that could later be discarded due to a lack of events required for the pre-event video sequence.

[0073] Alternatively, the image frames can be compressed before being stored in the pre-event buffer 112, such as with simple compression (e.g., lossless run-length encoding) using a dedicated hardware block compression unit 114, or with JPEG compression using, for example, 4×4 or 8×8 blocks (which consumes little power and still reduces the size of captured image frames by about ten times). The chosen compression method is typically of the type that avoids time dependencies between images. In this way, the same number of image frames can be stored using a smaller pre-event buffer.

[0074] Furthermore, or alternatively, limited image processing can be performed with very low power consumption processing steps before storing the image frame in the pre-event buffer 112. This can be illustrated by moving such steps from the image processing unit 106 to a separate initial image processing unit 116 and allowing processing of the image frame therebefore storing it in the pre-event buffer 112. It can be noted that it is illustrated as a separate unit merely for illustrative purposes, and it can also be illustrated as having an output that feeds the image to the pre-event buffer 112 after only performing such initial steps as illustrated here.

[0075] Examples of steps that the initial image processing unit 116 can perform before being stored in the pre-event buffer 112 are defect pixel correction and demosaicing, the latter also known as color interpolation or reconstruction, which is the process of converting the image from Bayer values ​​to RGB values. The reason those steps are performed in the initial image processing unit 116 is that they consume a limited amount of power and they are performed at the start of image processing.

[0076] Furthermore, the Bayer-to-RGB conversion results in increased smoothness, which in turn makes the image more compressible. This allows the optional compression unit 114 to increase the reduction in image size, which in turn enables the use of a smaller pre-event buffer 112 to store a collection of pre-event image frames, or, if desired, allows a large number of pre-event image frames to be stored in the pre-event buffer 112.

[0077] However, it is important to note that the pre-event image frames are stored in the pre-event buffer 112 in a largely unprocessed or raw format to minimize power consumption when processing images that may not be of interest and will be discarded. It should be noted that the camera 100 continuously captures video of the scene, but the images are not fed to the image processing unit 106 and the memory 108. As described above, the captured image frames are typically stored temporarily only in the pre-event buffer 112.

[0078] Therefore, image frames are not sent to the image processing unit 106 until an event occurs that requires the camera 100 to record video. However, as will be discussed in more detail below, there may be a situation where a subset of image data from the image capture unit 104 is also fed to the image processing unit 106 during the period when no event occurs.

[0079] exist Figure 2 The image shows a typical scenario of camera 100 in use. Camera 100 is used to monitor scene 200 and is worn by person 202. Various activities can occur in scene 200, and at a certain point in time, person 202 can determine that an event is occurring that requires video recording to begin.

[0080] Person 202 then activates a switch or presses button 204 on camera 100 to allow video recording to begin. When button 204 is pressed, a first signal 120 is sent to event unit 118 located in camera 100. Pressing button 204 is one of several examples in which an external event detection unit sends a signal to camera 100 (or more precisely, to event unit 118) that an event requiring video recording has occurred, started, or is occurring.

[0081] Such an external event detection unit, and other examples of events detected that could lead to the first signal 120, include pressure sensors or motion sensors that signal that a weapon has been removed from its holster (typically in the case where camera 100 is carried by a police officer), or sensors including accelerometers or gyroscopes that signal that a specific type of movement is occurring, indicating that the person carrying camera 100 is moving in a specific manner. One example of this is a movement indicating that a fall has occurred, such as when the camera (and the person) has moved from a vertical position (standing) to a horizontal position (lying on the ground). Another example is a sensor detecting a movement indicating that the person carrying the camera is being pushed or shoved.

[0082] Another example of an external event detection unit that can provide the first signal 120 to event unit 118 is a motion detector that detects motion in a scene. The motion detector can be a PIR sensor, a depth sensor, a laser, radar, or another type of device capable of sensing motion in a scene. More advanced analysis of the scene can also be performed, which may indicate the probability of ignoring motion below a threshold, repetitive motion, or motion from a specific scene area known to include, for example, swaying trees. All of these are examples that could cause the external event detection unit to send the first signal 120 to event unit 118.

[0083] Another example of an external event detection unit that generates the first signal would be an audio analysis unit, which is configured to capture and analyze sounds or audio in the scene (such as gunshots or screams indicating that an event is taking place), and then send the first signal 120 in response to this.

[0084] Another example of an external event detection unit is a location determination unit, which uses GPS signals or some other positioning signal to determine that the camera has arrived at the location or area where video needs to be recorded. In response to such determination, the external event detection unit sends a first signal 120 to the event unit 118. A change in connection status to a GPS network or another type of positioning network can also cause the first signal 120 to be sent.

[0085] Other examples of events that can be detected by an external event detection unit and cause the first signal 120 to be sent to the event unit 118 are changes in the connection to a communication network (such as a connection to a Wi-Fi network or a mobile phone network).

[0086] Another example of how the first signal can be implemented is by using an image analysis unit 122, which analyzes the captured image frames to find specific content in the scene (such as motion with specific characteristics, or objects that meet specific predetermined criteria).

[0087] When event unit 118 receives the first signal 120, image capture unit 104 will stop feeding image frames to the pre-event buffer 112, and newly captured image frames will instead be fed by the image capture unit to the image processing unit 106 that performs image processing on the image frames. The processed image frames from image processing unit 106 are stored in memory 108 as a post-event video sequence.

[0088] At some point after event unit 118 receives the first signal, event unit 118 will receive a second signal 124. The second signal 124 indicates that the event requiring video recording has stopped or (in some cases) may have stopped. The second signal can be generated by person 202 pressing button 204 a second time to indicate that video recording can be stopped. A second button can also be provided on camera 100 for the purpose of generating the second signal.

[0089] Another option for generating the second signal 124 is a timer that starts upon receiving the first signal 120 and generates the second signal 124 after a predetermined time has elapsed, or after input from a sensor included in the external event detection unit, or after no input from a sensor included in the external event detection unit. For example, the second signal 124 could be generated by a sensor indicating that a weapon has been reinserted into its holster or by a sensor indicating that movement has stopped. Furthermore, as when the first signal is generated, the image analysis unit 122 can analyze image frames to determine that the event triggering video recording has now stopped, thereby triggering the generation of the second signal.

[0090] Other types of input can also be used to generate a second signal. Such input can come from sensors that indicate, for example, that a person carrying a camera has left a specific area or location, that sound has stopped, or that movement in the scene has stopped or an object has left the scene.

[0091] As previously mentioned, it is often necessary to check what happened before the first signal 120 was generated; that is, a pre-event video sequence is required. Therefore, upon receiving the second signal 124, the feeding of image frames from the image capture unit 104 to the image processing unit 106 will be suspended. Instead, when the image processing unit 106 receives image frames from the image capture unit 104, the set of image frames waiting in the pre-event buffer 112 will now be fed to the image processing unit 106. There, the set of images will be processed, and then the processed set of images will be added to the memory 108 as a pre-event video sequence for the post-event video sequence previously stored in the memory 108. Furthermore, the image capture unit 104 will resume feeding newly captured images to the pre-event buffer 112, which is now ready to resume receiving images.

[0092] It can be noted that a set of image frames from the pre-event buffer can be processed at a rate lower than the full frame rate; that is, more time can be used to process each image frame than when processing real-time video. In this way, more advanced image processing can be used, or the image processing unit 106 can be configured to process each image frame using less power (e.g., by configuring it to operate at a lower clock frequency). Processing the set of image frames at a rate lower than the full frame rate is advantageously combined with using a larger pre-event buffer to allow new, potential pre-event image frames to be stored while the set of image frames is being processed.

[0093] exist Figure 3 The diagram illustrates a method 300 for recording an event video sequence. In step 302, image capture unit 104 captures image frames. In optional step 304, the image frames are compressed before being stored in pre-event buffer 112 in step 306. At some point, in step 308, a first signal 120 indicating that an event has occurred is received. Then, image capture unit 104 stops sending image frames to pre-event buffer 112 for storage, and instead sends image frames to image processing unit 106 for processing and then to memory 108 for storage, to generate a post-event video sequence in step 310.

[0094] As an optional measure in this step, the processed image may also be streamed to a remote receiver (e.g., in the form of a server), which, upon receiving the first signal 120, provides the logged-in operator with immediate access to the video captured by the camera 100 so that the operator can determine whether action is needed (such as sending assistance to the person 202 carrying the camera 100).

[0095] In step 312, a second signal 124 indicating the end of the event is received. Then, in step 314, a pre-event video sequence is generated by feeding a set of image frames stored in the pre-event buffer 112 to the image processing unit 106 for processing and by feeding the processed set of images to the memory 108 for storage as a pre-event video sequence. If the image frames are compressed in relation to those stored in the pre-event buffer 112, appropriate decompression is performed before processing in the image processing unit 106. In step 316, the post-event video sequence and the pre-event video sequence together form an event video sequence stored in the memory 108.

[0096] It should be noted that step 302 occurs continuously during camera operation and is presented as the first step only for illustrative purposes. Furthermore, upon receiving the second signal, when the contents of the pre-event buffer 112 are fed to the image processing unit, the image capture unit 104 again feeds image frames to the pre-event buffer 112. After optionally compressing the image, the pre-event buffer 112 stores the image frames as a new set of image frames in preparation for the occurrence of the next event and the generation of a new pre-event video sequence.

[0097] As previously described, in a variation of the solution proposed herein, during the time it takes for image frames to be fed to the event pre-buffer 112, i.e., when no event is indicated to have occurred, a subset of image data can be fed from the image capture unit 104 to the image processing unit 106. This subset of image data can, for example, be used for streaming from the camera 100. The subset of image data can be provided by spatially subsampling the captured image frames to provide lower resolution image frames, or by temporally subsampling the captured image frames to provide a reduced frame rate, or both. The processed subset of image data can also be stored in the memory 108 and used when generating an event video sequence.

[0098] The storage of a processed subset of image data can occur in a manner similar to how images are stored in the pre-event buffer 112; that is, older image frames may be discarded after a certain amount of time has elapsed or after a certain number of new image frames have been added to memory. The camera can be configured to feed only such image data from the image capture unit 104 to the pre-event buffer 112, image data that has not yet been included in the temporal or spatial subsampling image data used for streaming. When a pre-event video sequence is generated, the image processing unit 106 then combines the set of image frames from the pre-event buffer 112 with the previously processed subset of image data to provide a complete pre-event video sequence.

[0099] It should be noted that the different units presented in the above text (e.g., image processing unit 106, compression unit 114, initial image processing unit 116, event unit 118, and image analysis unit 122) can all be implemented in software, hardware, or any combination thereof. One or more units can be provided in the form of dedicated circuits configured to perform the functions of the respective units, or as a set of software program instructions implemented to run on a standard processing unit and perform the functions of the respective units.

[0100] Finally, it should be mentioned that those skilled in the art, when practicing the claimed invention, can understand and implement other variations of the disclosed embodiments by studying the drawings, the disclosure, and the appended claims.

[0101] One example of such a variation is that, even though the invention has been described in conjunction with a wearable camera, it can also be used in cameras, for example, mounted on buildings or in vehicles. If the camera has limited available power, or if there is a desire to conserve power related to event video recording for other reasons, such as the availability of more ample storage space in a pre-event buffer, then the solutions described herein can be used to provide reduced power consumption during those periods when no events are occurring.

[0102] In summary, a method is proposed to provide a video sequence initiating an event, comprising a pre-event video sequence and a post-event video sequence. Pre-event image frames are stored in an unprocessed format until the event occurs. Power saving is achieved by delaying image processing of the pre-event video sequence.

Claims

1. A method for recording a video sequence of events, wherein, The event video sequence includes: Pre-event video sequences show the monitored scene during a predetermined time period prior to event detection; and The video sequence following the event shows the monitored scene starting from the time the event was detected. The method includes the following steps: Capture image frames continuously; The newly captured image frames are continuously stored in the pre-event buffer without being processed in the image processing unit, by adding the newly captured image frames to the set of pre-event image frames corresponding to the predetermined time length. Receive the first signal indicating that an event has been detected; After receiving the first signal, the post-event video sequence is generated through the following steps: The step of storing the newly captured image frame in the pre-event buffer is aborted. The newly captured image frame is fed to the image processing unit. Image processing of the newly captured image frame is performed in the image processing unit, and The processed image frames are stored in memory as a post-event video sequence. Receive a second signal indicating that the event has ended; Upon receiving the second signal, the pre-event video sequence is generated through the following steps: Stop feeding the newly captured image frame to the image processing unit. The set of image frames is fed from the pre-event buffer to the image processing unit. Image processing of the set of image frames is performed in the image processing unit, and The processed set of image frames is stored in the memory as the pre-event video sequence. The image processing steps include one or more of the following: defect pixel correction, white balance, demosaic, matrixing, gamma correction, sharpening, noise filtering, scaling, and encoding.

2. The method of claim 1, further comprising the step of continuously discarding the oldest image frames from the pre-event buffer after storing newly captured image frames.

3. The method of claim 1, further comprising feeding a subset of image data of the newly captured image frame to the image processing unit before receiving the first signal.

4. The method of claim 3, wherein, A subset of the image data is generated by spatially subsampling the newly captured image frame, such that image frames with reduced resolution compared to the newly captured image frame are fed to the image processing unit.

5. The method of claim 3, wherein, A subset of the image data is generated by temporally subsampling the newly captured image frames, such that an image frame stream with a reduced frame rate compared to the frame rate of the newly captured image frames is fed to the image processing unit.

6. The method of claim 1, further comprising, after feeding the set of image frames from the pre-event buffer to the image processing unit, restoring the continuous storage of the newly captured image frames in the pre-event buffer.

7. The method of claim 1, wherein, The first signal is received from an image analysis unit, which is configured to detect events in the monitored scene based on captured image frames.

8. The method of claim 1, wherein, The first signal is received from an external event detection unit, which is configured to detect events in the monitored scene.

9. The method of claim 1, wherein, The second signal is generated as a response to no longer receiving the first signal.

10. The method of claim 1, wherein, The second signal is generated in response to the timer reaching a predetermined value, and the timer is started when the first signal is received.

11. The method of claim 1, further comprising, upon receiving the first signal indicating that an event has been detected, streaming the processed image frames of the post-event video sequence to a receiver.

12. A surveillance camera configured to record video sequences of events, in, The event video sequence includes: The pre-event video sequence contains video showing the monitored scene during a predetermined time period prior to event detection; and The post-event video sequence contains videos showing the monitored scene starting from the time the event was detected. The camera includes: The image capture unit is configured to continuously capture image frames; The pre-event buffer is configured to continuously store the newly captured image frames by adding them to a set of pre-event image frames corresponding to the predetermined time length without processing the newly captured image frames in the image processing unit. The event unit is configured to receive a first signal indicating that an event has been detected and a second signal indicating that the event has ended; The image processing unit is configured to process image frames; The memory is configured to store processed image frames. Upon receiving the first signal, the post-event video sequence is generated through the following steps: The camera is configured to stop storing the newly captured image frame in the pre-event buffer and is configured to feed the newly captured image frame to the image processing unit. The image processing unit is configured to perform image processing on the newly captured image frame and store the processed image frame as the post-event video sequence in the memory. Upon receiving the second signal, the video sequence preceding the event is generated through the following steps: The camera is configured to stop feeding newly captured image frames to the image processing unit. The camera is configured to feed the set of image frames from the pre-event buffer to the image processing unit. The image processing unit is configured to perform image processing on the set of image frames. The camera is configured to store a set of processed image frames in the memory as a pre-event video sequence; The image processing includes one or more of the following: defect pixel correction, white balance, demosaic, matrixing, gamma correction, sharpening, noise filtering, scaling, and encoding.

Images