Method of controlling camera for monitoring scene

Abstract

A method of controlling a camera for monitoring a scene includes determining a first camera setting for monitoring the scene under a first lighting condition and a second camera setting for monitoring the scene under a second lighting condition; obtaining a reference image representing a scene captured with the camera set to the first camera setting under a second illumination condition; while monitoring the scene with the camera set as the first camera setting, detecting a change in the scene from the first illumination condition to the second illumination condition, the detecting the change comprising performing a comparison between: after the change from the first illumination condition to the second illumination condition, detecting the change from the first illumination condition to the second illumination condition; first image feature data derived from a first image of the scene captured with the camera set as the first camera setting, and reference image feature data derived from the reference image, to determine a match of the first image feature data to the reference image feature data; and in response to detecting the change, switching the camera from the first camera setting to the second camera setting and continuing to monitor the scene.

Description

Technical Field

[0001] The present invention generally relates to a method for controlling a camera for monitoring a scene. Background Art

[0002] Surveillance cameras used in monitoring applications typically employ exposure algorithms to automatically control the exposure-related settings of the camera to allow for the capture of images that are properly exposed under changing lighting conditions in the monitored scene. In scenes that are primarily illuminated by the sun, such as outdoor scenes, changes in lighting conditions tend to occur at a relatively slow pace, such that the exposure algorithm can fairly well track the changing lighting conditions in the scene.

[0003] However, some scenes, such as those located in indoor environments, may be dominated by light that can change rapidly and randomly. For example, indoor environments, such as rooms with only a few windows or no windows, are primarily illuminated by light sources that can switch between on and off and / or by indirect light (artificial light or sunlight) that can enter the room through doors that can be opened and closed. In such an environment, the lighting conditions can change from relatively dark (e.g., when the light source is turned off or the door is closed) to relatively bright (e.g., when the light source is turned on or the door is opened) in a short period of time, and vice versa. In the context of a monitoring application, the time required for an automatic exposure algorithm to adapt to the changed lighting conditions by incrementally adjusting the exposure-related settings of the camera (e.g., aperture, shutter speed, ISO value, etc.) can be long. In particular, during this time period, the images captured by the camera may be overexposed or underexposed, such that objects in the scene (e.g., a person turning the light source on or off or a person entering the room through the door) may not be easily distinguishable by an operator or by an image-based automatic object tracking algorithm. Further envision that, from a surveillance perspective, images captured in relation to such events may be particularly valuable, e.g., particularly valuable by having potentially high evidentiary value. Summary of the Invention

[0004] In view of the above, it is an object of the present invention to provide a method for controlling a camera for monitoring a scene, which method can adapt to changed lighting conditions more quickly than conventional automatic exposure algorithms. Another object of the present invention is to be able to more reliably monitor scenes in which lighting conditions may change rapidly and unpredictably. These and other objects can be better understood from the following.

[0005] Accordingly, in a first aspect of the present invention, there is provided a method for controlling a camera for monitoring a scene, the method comprising:

[0006] Determine a first camera setting for monitoring a scene under a first lighting condition and a second camera setting for monitoring the scene under a second lighting condition different from the first lighting condition;

[0007] Obtain a reference image, where the reference image represents the scene captured by a camera set to the first camera setting under the second lighting condition;

[0008] While monitoring the scene with a camera set to the first camera setting, detect a change of the scene from the first lighting condition to the second lighting condition, where detecting the change from the first lighting condition to the second lighting condition includes performing a comparison between:

[0009] First image feature data derived from a first image of the scene captured by a camera set to the first camera setting after changing from the first lighting condition to the second lighting condition, and

[0010] Reference image feature data derived from the reference image,

[0011] To determine a match between the first image feature data and the reference image feature data; and

[0012] In response to detecting the change, switch the camera from the first camera setting to the second camera setting and continue monitoring the scene.

[0013] By determining a first camera setting (which is a camera setting suitable for monitoring the scene under the first lighting condition) and a second camera setting (which is a camera setting suitable for monitoring the scene under the second lighting condition), and obtaining a reference image representing the scene captured by a camera set to the first camera setting and thus a "wrong" camera setting under the second lighting condition, a change from the first lighting condition to the second lighting condition can be detected quickly and reliably. In response, the camera setting can be switched directly from the first camera setting to the second camera setting without relying on the iterative and progressive changes performed by traditional automatic exposure algorithms.

[0014] Thus, the reference image (also referred to hereinafter as the "first reference image") can be considered in a sense as a "shortcut image", providing a shortcut from the first camera setting to the second camera setting.

[0015] The method of the first aspect includes making an image feature-based comparison between a reference (shortcut) image and a first image captured after changing from a first illumination condition to a second illumination condition. Since the reference image represents the scene captured under the second illumination condition using the first camera settings, the first image feature data (derived or extracted from the first image) can closely match the reference image feature data (derived or extracted from the reference image) when the second illumination condition occurs. This helps to reliably detect whether a switch change from the first camera settings to the second camera settings has been triggered while reducing the risk of switching to the second camera settings when the second illumination condition does not exist. That is, the method can reduce the risk of confusing another illumination condition different from the first and second illumination conditions with the second illumination condition and thus inadvertently triggering a change to the second camera settings (which may not be suitable for that other illumination condition).

[0016] It should be noted here that the "first" in the term "first image" is used as a label for the image captured after changing from the first illumination condition to the second illumination condition and should not be construed as the "first image" necessarily being the first image in the temporal sequence of images captured after changing from the first illumination condition to the second illumination condition. That is, if the first image is indeed the first image in the temporal sequence of images captured after changing from the first illumination condition to the second illumination condition, the changed illumination condition can be detected with a smaller delay and thus the camera settings can be switched faster.

[0017] The first image can be a frame in a frame sequence captured by a camera set to the first capture settings during a surveillance scene. In particular, the frame sequence can be a video sequence.

[0018] The reference image can generally be the scene captured under the second illumination condition by a camera set to the first camera settings. Thus, when set to the first camera settings, the reference image can accurately reflect the scene captured by the camera under the second illumination condition.

[0019] In some embodiments, the method further includes: deriving reference image feature data and storing the reference image feature data, wherein performing the comparison includes: comparing the first image feature data with the stored reference image feature data. Thus, the reference image feature data can be derived and stored once instead of being derived each time it is compared with a captured image.

[0020] In some embodiments, the first image feature data and the reference image feature data include exposure-related data. In terms of the comparison for the purpose of detecting the changed illumination condition, the exposure-related data can be a particularly relevant metric.

[0021] The exposure-related data of the first image feature data and the reference image feature data may each include statistical data of pixel values respectively derived from the first image and the reference image. The statistical data may include one or more of the following: representative pixel values such as average pixel value or median pixel value for the first image or the reference image, contrast, frequency distribution of pixel values. Any of these statistical data, either individually or in combination, may contribute to detecting a changed illumination condition in the scene.

[0022] In some embodiments, the first image feature data and the reference image feature data include pixel values of spatially corresponding pixels of the first image and the reference image. Thus, the comparison of the first image feature data with the reference image feature data may be equivalent to comparing the pixel values of the spatially corresponding pixels derived from the first image and the reference image.

[0023] In some embodiments, the first image feature data and the reference image feature data are respectively derived from the first image and the reference image by downsampling the respective images and extracting the respective image feature data from the respective downsampled images. This reduces the amount of data that needs to be processed and stored and thus facilitates a computationally efficient implementation.

[0024] In some embodiments, the scene includes one or more regions of interest, each region of interest being depicted in a respective one of one or more portions in the first image and the reference image, and wherein the first image feature data and the reference image feature data are respectively derived from at least one or more portions in the first image and the reference image.

[0025] By detecting based on the image feature data derived from the portions of the first image and the reference image that depict one or more regions of interest, the detection can take into account the relatively more interesting portions of the image in a given surveillance scenario. The "regions of interest" of the scene and the corresponding "portions" of the first image and the reference image are here understood as (strict) sub-regions of the scene and (strict) subsets of the first image and the reference image, respectively.

[0026] For example, the first image feature data and the reference image feature data may include exposure-related data, wherein the exposure-related data of the first image feature data and the exposure-related data of the reference image feature data may include statistical data of pixel values respectively derived from corresponding and spatially corresponding portions of the first image and the reference image (i.e., one or more portions that depict one or more regions of interest in the scene). Thus, the comparison of the first image feature data and the reference image feature data may include comparing the statistical data of pixel values derived from the spatially corresponding portions of the first image and the reference image that depict the regions of interest of the scene.

[0027] The further effect of considering one or more regions of interest to detect a change from a first lighting condition to a second lighting condition is that, in many scenarios, the second lighting condition does not need to uniformly affect the illuminance level throughout the scene. Thus, the image feature data can be derived from one or more image portions where the changed illuminance level may affect the reliability of the surveillance.

[0028] The first image feature data and the reference image feature data can be derived respectively only from one or more portions of the first image and the reference image. That is, portions of the first image and the reference image that do not depict the regions of interest can be excluded from the image feature data as well as the comparison. That is, in order to detect the changed lighting condition, regions of the scene that are not of interest in a given surveillance scenario can be ignored.

[0029] One or more regions of interest can include objects such as doorways or artificial lighting that have a first state and a second state (e.g., a lower brightness state and a higher brightness state), where the first state is associated with the first lighting condition and the second state is associated with the second lighting condition. Doorways as well as artificial lighting both represent typical regions of interest in the scene that may suddenly change their brightness state and thus cause a change in the lighting condition of the scene. From a surveillance perspective, the door may be more interesting because it is the location where people can enter and exit the scene.

[0030] In some embodiments, the first lighting condition and the second lighting condition can respectively correspond to a first steady-state lighting condition and a second steady-state lighting condition having different brightness levels. Correspondingly, the first camera settings and the second camera settings can be the respective settings of the camera that are suitable for the first steady-state lighting condition and the second steady-state lighting condition. Thus, the first camera settings and the second camera settings can be their respective "static settings" or "static modes", representing relatively stable or unchanging settings of the camera. In the example related to the indoor environment discussed in the background art, the first steady-state lighting condition can correspond to the light source being turned off or the door being closed, while the second steady-state lighting condition may correspond to the light source being turned on or the door being opened, or vice versa.

[0031] In some embodiments, the first camera settings and the second camera settings include settings of one or more exposure-related control parameters of the camera.

[0032] The first camera settings and the second camera settings can each include settings of one or more of the following: shutter speed, aperture, ISO value, camera illumination (visible light or infrared (IR)), and IR filter state. Any one of these exposure-related control parameters of the camera, either individually or in combination, allows controlling the exposure level when capturing an image.

[0033] In some embodiments, the method further comprises:

[0034] obtaining a second reference image, wherein the second reference image represents a scene captured by a camera set to a second camera setting under a first illumination condition;

[0035] while monitoring the scene with a camera set to a second camera setting, detecting a change of the scene from a second illumination condition to the first illumination condition, wherein detecting the change from the second illumination condition to the first illumination condition comprises performing a comparison between:

[0036] second image feature data derived from a second image of the scene captured by a camera set to a second camera setting after changing from the second illumination condition to the first illumination condition, and

[0037] second reference image feature data derived from the second reference image,

[0038] to determine a match between the second image feature data and the second reference image feature data; and

[0039] in response to detecting the change, switching the camera from the second camera setting to the first camera setting and continuing to monitor the scene.

[0040] Thus, the second reference or "snap" image can be used to quickly and reliably detect a change from a second illumination condition to the first illumination condition in a corresponding manner, and in response, directly switch from the second camera setting to the first camera setting, without relying on the iterative and gradual changes performed by conventional automatic exposure algorithms.

[0041] The above discussion of the first image feature data and the (first) reference image feature data correspondingly applies to the second image feature data and the second reference image feature data.

[0042] The second reference image can generally be a scene captured by a camera set to a second camera setting under a first illumination condition. When set to the second camera setting, the second reference image can accurately reflect the scene captured by the camera under the first illumination condition.

[0043] According to a second aspect of the present invention, there is provided a camera comprising: processing means configured to perform the method of the first aspect, or any embodiment or example thereof.

[0044] According to a third aspect of the present invention, there is provided a computer program product comprising a computer program code portion configured to, when executed by processing means, perform a method of controlling a camera for monitoring a scene, the method comprising:

[0045] Determine a first camera setting for monitoring the scene under a first lighting condition and a second camera setting for monitoring the scene under a second lighting condition different from the first lighting condition;

[0046] Obtain a reference image, where the reference image represents the scene captured by a camera set to the first camera setting under the second lighting condition;

[0047] While the camera is set to the first camera setting and monitoring the scene, detect a change of the scene from the first lighting condition to the second lighting condition, where detecting the change from the first lighting condition to the second lighting condition includes performing a comparison between:

[0048] First image feature data derived from a first image of the scene captured by a camera set to the first camera setting after changing from the first lighting condition to the second lighting condition, and

[0049] Reference image feature data derived from the reference image,

[0050] To determine a match between the first image feature data and the reference image feature data; and

[0051] In response to detecting the change, switch the camera from the first camera setting to the second camera setting and continue monitoring the scene.

[0052] The features of the second and third aspects have the same or equivalent advantages as the first aspect. Any feature described with respect to the first aspect may have a corresponding feature in the second and third aspects, and vice versa. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] This aspect of the invention and other aspects will now be described in more detail with reference to the accompanying drawings that illustrate embodiments of the invention.

[0054] Figure 1 A block diagram schematically showing an implementation of a camera is shown.

[0055] Figure 2 A first example scene is shown.

[0056] Figure 3 A second example scene is shown.

[0057] Figure 4 Is a flowchart of a method for controlling a camera for monitoring a scene, the method including detecting a change from a first lighting condition to a second lighting condition.

[0058] Figure 5 Is a flowchart of method steps for detecting a change in lighting condition in a scene, the method steps being part of Figure 4 the method of.

[0059] Figure 6 is a flowchart of a method for controlling a camera for monitoring a scene, the method including detecting a change from a second lighting condition to a first lighting condition.

[0060] Figure 7 shows a schematic example of an image depicting Figure 2 a first example scene of

[0061] Figure 8 shows a schematic example of an image depicting Figure 3 a second example scene of DETAILED DESCRIPTION

[0062] Figure 1 is a schematic block diagram of a camera 20 that can implement the method of the present disclosure. The camera 20 can be a surveillance camera or a monitoring camera. The camera 20 includes an image capture device 22 that includes, for example, an image sensor and optics. The camera 20 monitors a scene 10 by capturing images or frames using the image capture device 22 to provide an image sequence or frame sequence 28 of the scene 10. The images 28 can be captured at a predetermined frame rate or a variable frame rate suitable for a given surveillance application. In particular, the image sequence 28 can form the frames of a video sequence of the scene 10.

[0063] The camera 20 further includes a processing device 24 and a memory 26. The memory 26 can be associated with the processing device 24, for example, by being coupled to or included in the processing device 24. The processing device 24 is configured to receive the captured image sequence 28 and process it. As Figure 1 indicated in

[0064] As described below, the method for controlling camera 20 can be implemented in hardware and software. In a software implementation, processing device 24 can be implemented in the form of one or more processors (such as one or more central processing units), which are associated with computer program code instructions stored on a (non-transitory) computer-readable medium (such as non-volatile memory), causing processing device 24 to execute the method steps for controlling camera 20. Examples of non-volatile memory include read-only memory, flash memory, ferroelectric RAM, magnetic computer storage devices, and optical discs, etc. In a hardware implementation, processing device 24 can instead be implemented by dedicated circuitry configured to implement the method steps for controlling camera 20. The circuitry can be in the form of one or more integrated circuits, such as one or more application-specific integrated circuits (ASICs) or one or more field-programmable gate arrays (FPGAs). It should be understood that a combination of hardware implementation and software implementation can also be had, which means that some method steps can be implemented in dedicated circuitry and other steps can be implemented in software.

[0065] Figure 2 An example indoor space in the form of room 1 is shown. Camera 20 is installed (i.e., arranged) in room 1 to monitor scene 11 of room 1. As indicated by the dashed line extending from camera 20, scene 11 corresponds to the portion of room 1 within the field of view of camera 20. Room 1 includes door 2 disposed at the doorway connecting room 1 to adjacent space 3. Adjacent space 3 can be an adjacent room (e.g., a corridor) or an outdoor space. Assuming room 1 is dark, dimly illuminated only by artificial light sources or illuminated by stray light entering room 1 through one or more small windows of room 1, the lighting conditions of scene 11 may be relatively dark or dim when door 2 is closed. Further assuming adjacent space 3 is brightly illuminated (e.g., by sunlight or by bright artificial lighting in space 3), the lighting conditions of scene 11 may be relatively bright when door 2 is opened.

[0066] Figure 3 Another example indoor space in the form of room 1' is shown. Reference numeral 12 denotes the scene in room 1' monitored by camera 20. Figure 3 Differing from Figure 2 is that it is not assumed that adjacent space 3 is brightly illuminated. Thus, it is assumed that the state of door 2 (i.e., open or closed) does not have any substantial effect on the illuminance levels in room 1' and scene 12. Instead, room 1' can be mainly illuminated by artificial lighting 4 (e.g., including one or more lighting devices), which can be switched on or off, for example, by a light switch disposed inside or outside room 1'. Thus, the lighting conditions of scene 12 may be relatively dark or dim when artificial lighting 4 is switched off and relatively bright when artificial lighting 4 is switched on.

[0067] Accordingly, Scenarios 11 and 12 are examples of scenarios in which the first lighting condition or the second lighting condition may exist. The first lighting condition may correspond to a lower illuminance level in Scenarios 11 and 12, and the second lighting condition may correspond to a higher illuminance level in Scenarios 11 and 12, or vice versa. In Figure 2 and Figure 3 , the first lighting condition and the second lighting condition may correspond to a first steady-state lighting condition and a second steady-state lighting condition having different illuminance levels, respectively. That is, when the door 2 is closed or the artificial lighting 4 is turned off, the illuminance levels in the respective Scenarios 11 and 12 may be relatively stable, and the same applies when the door 2 is opened or the artificial lighting 4 is turned on.

[0068] In addition, the door 2 and the artificial lighting 4 are examples of respective objects in the monitored Scenarios 11 and 12, having a first state and a second state (e.g., open / closed, on / off) associated with different lighting conditions in the scenario. Accordingly, when the door is in the closed state, the first lighting condition may exist in Scenario 11, and when the door 2 is in the open state, the second lighting condition may exist in Scenario 11, or vice versa. Correspondingly, when the artificial lighting 4 is in the off state, the first lighting condition may exist in Scenario 12, and when the artificial lighting 4 is in the on state, the second lighting condition may exist in Scenario 12, or vice versa.

[0069] A method of controlling the camera 20 as shown in Figures 4 to 5 will now be described with reference to the flowchart of Figure 1 . The scenario 10 monitored by the camera 20 may be, for example, Figures 2 to 3 either of Scenarios 11 and 12 of

[0070] At S10, the processing device 24 of the camera 20 determines a first camera setting suitable for monitoring the scenario 10 under the first lighting condition, and at S11, determines a second camera setting suitable for monitoring the scenario 10 under a second lighting condition different from the first lighting condition. As shown in Figure 1 , the processing device 24 may store the first camera setting 30 and the second camera setting 32 in the memory 26 for later retrieval. For simplicity, it will be assumed hereinafter that the first lighting condition corresponds to a lower illuminance level in the scenario 10, and the second lighting condition corresponds to a higher illuminance level in the scenario 10. However, the reverse case may also apply equally well.

[0071] When the scene 10 is under the first lighting condition and the second lighting condition respectively, an automatic exposure algorithm can be used to determine the first camera settings and the second camera settings. The automatic exposure algorithm can be implemented in a conventional manner. For example, the automatic exposure algorithm can be configured to control the settings of one or more exposure-related control parameters of the camera 20 to obtain a target exposure. Examples of exposure-related control parameters include shutter speed, aperture, ISO value, camera illumination (visible light or infrared light), and infrared filter state. The automatic exposure algorithm can be configured to determine the current exposure level for the scene 10 and (e.g., incrementally and / or step by step) adjust the settings of one or more exposure-related control parameters to reduce the difference between a predetermined target exposure level and the current exposure level.

[0072] The first camera settings and the second camera settings can be determined as part of the configuration steps or calibration steps of the method. That is, the scene 10 can be intentionally made to present the first lighting condition (e.g., by closing the door 2 or cutting off the artificial lighting 4), where the processing device 24 can use the automatic exposure algorithm to determine the first camera settings suitable for the first lighting condition. The scene 10 can be further made to present the second lighting condition (e.g., by opening the door 2 or turning on the artificial lighting 4), where the processing device 24 can use the automatic exposure algorithm to determine the second camera settings suitable for the second lighting condition.

[0073] The first camera settings and the second camera settings can also be determined during the monitoring of the scene 10. That is, the processing device 24 can record the frequency distribution of the camera settings determined by the automatic exposure algorithm over time (e.g., hours, days, or months). For a scene with a first steady-state lighting condition and a second steady-state lighting condition at different illuminance levels, such as scene 11 and scene 12, it can be expected that two different camera settings or two different sets of similar camera settings can be identified in this frequency distribution. Accordingly, the respective averages of the two different camera settings or multiple sets of similar camera settings can be determined as the first camera settings and the second camera settings.

[0074] Although S10 and S11 are shown as subsequent steps, it should be noted that S10 and S11 can be performed in any order. In the case of determining the first camera settings and the second camera settings over time, steps S10 and S11 can even be performed in an interleaved or interwoven manner.

[0075] At S12, the processing device 24 obtains a (first) reference image of the scene 10. This reference image can generally be when the scene 10 is under the second lighting condition (e.g., Figure 2 the door 2 in Figure 3when the artificial lighting 4 in is turned on) by the camera 20 set to the first camera setting (more specifically, by the image capture device 22 of the camera 20). That is, the reference image can be captured by the camera 20 using the first camera setting when the scene 10 is under the second lighting condition. Therefore, when set to the first camera setting, the reference image can accurately reflect the scene 10 seen or imaged by the camera 20 under the second lighting condition.

[0076] Similar to the first camera setting and the second camera setting, the reference image can be obtained as part of the configuration step or calibration step of the method. That is, the scene 10 can be intentionally made to present the second lighting condition, the camera 20 can be set to the first camera setting (thereby overriding the automatic exposure algorithm of the camera 20), and an image of the scene 10 can be captured by the camera 20. This image can be determined as the reference image. Multiple images of the scene 10 can also be captured using the first camera setting, and the reference image can be determined as the average image of these multiple images. It is further contemplated that the reference image can alternatively be obtained as an image of the scene 10 captured by another but similar camera (e.g., by a camera installer when deploying a camera for monitoring the scene 10) using the corresponding camera setting and subsequently transmitted to the memory 26 of the camera 20.

[0077] As will be further described below and as Figure 1 shown in , the reference image (which can be downsampled) 34a and / or the image feature data 34b derived from the (downsampled) reference image can be stored in the memory 26 as reference data 34 for subsequent use in the method.

[0078] At S13, the camera 20 continues to monitor the scene 10. During the monitoring, the camera 20 captures a temporal image sequence 28 of the scene 10 through the image capture device 22. Here it is assumed that the scene 10 is under the first lighting condition (e.g., Figure 2 the door 2 in is closed or Figure 3 the artificial lighting 4 in is turned off, making the scene 11 or 12 relatively dim or dark), and the camera 20 is set to the first camera setting (i.e., according to the first camera setting 30 stored in the memory 26). Therefore, at this stage, the captured images 28 of the scene 10 can be properly exposed. During the monitoring, the processing device 24 determines whether a change from the first lighting condition to the second lighting condition has occurred in the scene 10. This determination can be made at the same rate as the frame rate at which the images 28 are captured to allow for rapid detection of the change.

[0079] In response to not detecting a change from the first lighting condition to the second lighting condition, the monitoring can continue using the first camera setting.

[0080] In response to detecting a change from a first lighting condition to a second lighting condition, at S14, the processing device 24 switches the camera 20 from a first camera setting to a second camera setting. The processing device 24 may retrieve the second camera setting 32 from the memory 26 and control or set each parameter of the camera 20 according to the corresponding parameters of the second camera setting 32.

[0081] At S15, after switching the camera 20 to the second camera setting, the camera 20 continues to monitor the scene 10. Since the scene 10 is now under the second lighting condition, the captured image 28 of the scene 10 can be properly exposed.

[0082] Figure 5 Method steps for detecting a changed lighting condition in a scene are shown, and these method steps can be performed as Figure 4 part of step S13. For each image of the image sequence 28 captured by the camera 20, the Figure 4 method steps can be iteratively performed.

[0083] At S131, the processing device 24 obtains a current image captured by the camera 20 set to the first camera setting.

[0084] At S132, the processing device 24 derives image feature data from the current image.

[0085] At S133, the processing device 24 compares the image feature data derived from the current image (which may be interchangeably referred to as "current image feature data") with the (first) reference image feature data derived from a (first) reference image.

[0086] At S134, the processing device 24 determines whether the current image feature data matches the reference image feature data.

[0087] The reference image feature data may be pre-derived from a (scaled-down) reference image and stored in the memory 26 in association with obtaining the reference image as the reference image feature data 34b. In this case, the reference image feature data 34b can be retrieved from the memory 26 and compared with the current image feature data. In the case where a (scaled-down) reference image 34a (e.g., and not the reference image feature data) has been stored in the memory 26, the processing device 24 may derive the reference image feature data from the stored reference image 34a before making the comparison and then compare the derived reference image feature data with the current image feature data.

[0088] In response to determining that the current image feature data does not match the reference image feature data, the method proceeds according toFigure 5 the "mismatch" branch to continue execution, and thus, at S136, return to S131 to obtain the next image captured by camera 20 set to the first camera setting, and then use this next image as the current image to perform steps 132 to 134.

[0089] In response to determining that the current image feature data matches the reference image feature data, the method proceeds according to Figure 5 the "match" branch to continue execution, and thus, at S135, the processing device 24 determines that it has detected a change in scene 10 from the first lighting condition to the second lighting condition. Thereafter, the method proceeds to Figure 4 step S14 of to switch camera 20 to the second camera setting.

[0090] The comparison between the current image feature data and the reference image feature data may include: comparing the current image feature data and the reference image feature data to provide a similarity score (equivalently, a "match score"), and when this similarity score exceeds the similarity threshold, determining that the current image feature data matches the reference image feature data. The similarity threshold may be a predetermined threshold set to provide the desired tolerance for this comparison.

[0091] Refer to Figure 2 scene 11 of , as long as door 2 remains closed, and thus the first lighting condition is maintained in scene 11, then Figure 5 the method of can continue execution according to the "mismatch" branch. However, if door 2 is opened at some point during the monitoring, the lighting condition in scene 11 will change from the first lighting condition to the second lighting condition. Therefore, the image sequence 28 will include an image of scene 11 captured using the first camera setting under the first lighting condition (closed door 2). Refer to Figure 1 , this image may be, for example, image 281. The image sequence 28 will further include subsequent images of scene 11 captured using the first camera setting under the second lighting condition (open door 2). Therefore, this subsequent image is captured under the same lighting condition as the reference image and using the same camera setting, and thus will highly match the reference image, enabling the detection of the change from the first lighting condition to the second lighting condition. This subsequent image may be referred to as the "first image" and may be consistent with the previous discussion, and thus may be the first image in time captured after the change from the first lighting condition to the second lighting condition, or at least an image captured within a relatively short time window after the change from the first lighting condition to the second lighting condition (e.g., having a length corresponding to a relatively short frame sequence (such as within 20 frames or fewer, 10 frames or fewer, or 5 frames or fewer)) to reduce the latency in detecting the change to the second lighting condition. Refer toFigure 1 , the subsequent / first image can be, for example, the consecutive image 282 of image 281 or a subsequent image of image 281 (such as image 283 or 284).

[0092] Various types of image feature data can be derived and used for comparison. Hereinafter, the current image feature data and the reference image feature data will be referred to, where it should be understood that the current image feature data is related to the image feature data derived from the current image (such as the "first" image), and the reference image feature data is derived from the (first) reference image.

[0093] The current image feature data and the reference image feature data can include exposure-related data. The exposure-related data can include statistical data of pixel values respectively derived from the current image and the reference image. That is, the processing device 24 can process the current image and the reference image to derive the corresponding pixel value statistical data therefrom.

[0094] The processing device 24 can, for example, derive a representative pixel value for each of the current image or the reference image, such as an average pixel value or a median pixel value. Other examples include the maximum pixel value of each image or the mode (i.e., the most common) of the pixel values.

[0095] As another example, the processing device 24 can derive the corresponding contrast values of the current image and the reference image. Various contrast metrics are feasible, such as the contrast obtained by dividing the luminance difference (e.g., the difference between the maximum pixel value and the minimum pixel value of each image) by the average luminance of each image, or the root mean square (RMS) contrast of each image (e.g., corresponding to the standard deviation of the pixel values in each image).

[0096] As another example, the processing device 24 can derive the frequency distribution of the pixel values for each image. The "frequency distribution of pixel values" here refers to the statistical distribution indicating the number of occurrences of different pixel values (or pixel intensities) in each image. The frequency distribution can also be referred to as a histogram. The frequency distribution can indicate the absolute frequency or the relative frequency of the pixel values. The frequency distribution can be "binned", that is, it can indicate the frequency for multiple intervals (i.e., classes or sub-ranges) of pixel values defined within the pixel value range.

[0097] In the previous examples, the pixel value can generally refer to the luminance value of the pixel. However, in the case of a color image, the pixel value can also refer to the value of a color channel (e.g., any one of the RGB components of an RGB-encoded image or any one of the chrominance components of a YCbCr-encoded image), or an average (optionally weighted) combination of two or more components.

[0098] For any of these examples of exposure-related image feature data, the processing device 24 can compare the current image feature data with the reference image feature data to determine a similarity score. The similarity score can be determined by calculating the distance between the current image feature data and the reference image feature data (such as the distance between respective representative pixel values, contrast values, or frequency distributions). Any suitable conventional distance metric (such as the Euclidean distance) can be used to determine this distance. In the case of frequency distributions, the distance can be based on a differential frequency distribution determined by calculating the differences in pixel values between the frequency distributions. In response to the distance metric being less than a predetermined similarity threshold, it can be determined that the current image feature data matches the reference image feature data.

[0099] The current image feature data and the reference image feature data can optionally include the pixel values of spatially corresponding pixels of the current image and the reference image. Thus, the comparison of the current image feature data and the reference image feature data can be equivalent to a pixel-by-pixel comparison of the spatially corresponding pixels derived from the current image and the reference image to determine the distance between the pixels (such as the Euclidean distance). The similarity score can be determined as the mean squared error (MSE) between the pixel values. In response to the distance metric or the MSE being less than a predetermined similarity threshold, it can be determined that the current image feature data matches the reference image feature data.

[0100] The processing device 24 can determine one or more of the foregoing types of image feature data. If two or more types of image feature data are determined, the processing device 24 can determine a composite similarity score, for example, by determining the MSE based on the differences between the respective types of image feature data. In response to the distance metric or the MSE being less than a predetermined (combined) similarity threshold, it can be determined that the current image feature data matches the reference image feature data. When determining the composite similarity score, different weightings can be applied to different types of image feature data.

[0101] From a surveillance perspective, the scene 10 monitored by the camera 20 can include different regions of relative interest. For example, there may be little or no interest in certain regions of the scene, for example, because the region includes an impassable structure that prevents people from moving through the region, or because the region can only be accessed via another region that may be of higher interest. On the other hand, there may be relatively more interest in some regions, such as the entry points to an interior space (such as doorways) or regions that include inaccessible objects or structures. The latter type of region can be considered to represent the region of interest (ROI) of the scene.

[0102] Reference Figure 2, in scenario 11, the area 11a adjacent to the AND gate 2 can represent the ROI because this is the entry and exit point of room 1. Additionally, the area 11a can represent the ROI because it can be the area in scenario 11 where the greatest change in brightness level is expected when the gate 2 is opened or closed. If people open the gate 2 and enter room 1, the lighting conditions in scenario 11 will change from the first lighting condition to the second lighting condition. This may cause overexposure of the image captured immediately after the gate 2 is opened. This especially applies to the part of the captured image used to depict the ROI 11a, making it difficult to track and / or identify the people entering room 1.

[0103] Reference Figure 3 , in scenario 12, in addition to the first ROI 12a adjacent to the AND gate 2, it also includes a second ROI 12b surrounding the artificial lighting 4. The area 12b can represent the ROI because due to its proximity to the artificial lighting 4, this area 12b can be relatively well illuminated when the artificial lighting 4 is turned on. Additionally, the area 12b can represent the ROI because it may be desired to be able to detect attempts to tamper with the artificial lighting 4. Additionally, the lighting switch for the artificial lighting 4 may be located in the ROI 12b. Additionally, the area 12b can represent the ROI because it may be the area in scenario 11 where the greatest change in brightness level occurs when the artificial lighting 4 is turned on or activated. Therefore, in addition to the part used to depict the ROI 12a, it may also be desired to ensure proper exposure in the part of the captured image used to depict the ROI 12b.

[0104] The method for controlling the camera 20 can be adapted to take into account that scenario 10 may include one or more ROIs. Correspondingly, the method can further include determining, for each ROI in scenario 10, the part of each captured image used to depict the corresponding ROI. This part can be determined by obtaining pixel coordinates for the part used to depict each ROI. The pixel coordinates can be received as input by the installer during the installation of the camera 20, or after installation via the user interface of the camera monitoring software (e.g., via a network connected to the camera 20) as user input. The part of the image used to depict the ROI will hereinafter be referred to as the "ROI part".

[0105] Figure 7 is captured by the camera 20 Figure 2 Schematic example image 285 of scenario 11. Image 285 includes the ROI part 285a used to depict the ROI 11a including the gate 2. Figure 8 is captured by the camera 20 Figure 3Schematic example image 286 of scenario 12. Image 286 includes a first ROI portion 286a for depicting ROI 12a including door 2 and a second ROI portion 286b for depicting ROI 12b including artificial lighting 4. Any one of image 285 and image 286 represents any image in image sequence 28 captured by camera 20 during monitoring, as well as a reference image.

[0106] After determining the ROI portion (e.g., 285a or 286a to 286b), processing device 24 can derive current image feature data and reference image feature data from the ROI portions of the current image and the reference image respectively. For example, processing device 24 can determine corresponding exposure-related data for each ROI portion (e.g., 285a or 286a to 286b) of each of the current image and the reference image. For example, processing device 24 can determine corresponding representative pixel values, corresponding contrast values, and / or corresponding histograms for each ROI portion. Processing device 24 can also perform a pixel-by-pixel comparison of corresponding spatial pixels of the corresponding ROI portions of the current image and the reference image for each ROI portion. If the image includes two or more ROI portions (e.g., 286a to 286b), the combined similarity score can be determined based on the distance (e.g., difference) between the corresponding ROIs.

[0107] The change from the first lighting condition to the second lighting condition can be based only on the image feature data derived from the ROI portions, where the pixels of the non-ROI portions of the image can be excluded from the derivation and comparison of the image feature data. However, image feature data can also be derived from the non-ROI portions of the current image and the reference image and included in the common similarity score, but with a lower weight than the image feature data derived from the ROI portions.

[0108] In addition, regardless of whether the image feature data is derived from the entire image area or limited to one or more ROI portions, processing device 24 can derive the image feature data by extracting the image feature data from the full-resolution pixel data of the image. Alternatively, processing device 24 can generate downsampled representations of the current image and the reference image (e.g., thereby generating low-resolution "thumbnails"), and derive the image feature data from the downsampled representations. Various downsampling methods are feasible, such as subsampling, pre-filtering before subsampling to smooth or blur the image (e.g., using a Gaussian or other blur / smoothing kernel), interpolation (e.g., nearest neighbor, bilinear, bicubic), etc.

[0109] The change from the second lighting condition to the first lighting condition in scenario 10 can be detected and processed in a manner similar to the method for processing the first case described above. Figure 6It is a flowchart of a method for controlling camera 20 in response to detecting a change from a second lighting condition to a first lighting condition. Figure 6 The method steps of Figure 4 can be supplemented

[0110] At S21, the processing device 24 obtains a (second) reference image of scene 10. The second reference image can be obtained in a manner similar to the first reference image (i.e., the obtaining at S12). However, the second reference image instead represents scene 10 captured by the camera set to the second camera settings under the first lighting condition. Accordingly, the second reference image can generally be when scene 10 is under the first lighting condition (e.g., Figure 2 the door 2 in Figure 3 is closed or

[0111] the artificial lighting 4 in

[0112] is turned off) and is captured by camera 20 (more specifically, by the image capture device 22 of camera 20) set to the second camera settings. That is, the reference image can be captured by camera 20 using the second camera settings when scene 10 is under the first lighting condition. Thus, when set to the second camera settings, the second reference image can accurately reflect scene 10 as seen or imaged by camera 20 under the first lighting condition. Figure 1 As shown in

[0113] At S22, camera 20 continues to monitor scene 10. In particular, the method step S22 can be when scene 10 (e.g., Figure 2 scene 11 of Figure 3after the lighting condition in scenario 12) has changed from the first lighting condition to the second lighting condition and the camera 20 has been switched to the second camera setting. That is, method step S22 can be performed as a Figure 4 sub-step of S15.

[0114] During the monitoring, the processing device 24 captures the image sequence 28 and determines whether a change from the second lighting condition to the first lighting condition has occurred in scenario 10 (e.g., due to Figure 2 the closing of door 2 in Figure 3 or the cutting off of the artificial lighting 4 in

[0115] In response to not detecting a change from the second lighting condition to the first lighting condition, the monitoring can continue using the second camera setting.

[0116] In response to detecting a change from the second lighting condition to the first lighting condition, at S23, the processing device 24 switches the camera 20 from the second camera setting to the first camera setting. The processing device 24 can retrieve the first camera setting 30 from the memory 26 and control or set each parameter of the camera 20 according to the corresponding parameters of the first camera setting 30.

[0117] At S24, after switching the camera 20 to the first camera setting, the camera 20 continues to monitor scenario 10. Since scenario 10 is now under the first lighting condition, the captured images 28 of scenario 10 can be properly exposed.

[0118] By iteratively performing the following method steps by the processing device 24, a change from the second lighting condition to the first lighting condition can be detected in a manner similar to Figure 5 the flowchart of

[0119] a) Obtain the current image captured by the camera 20 set to the second camera setting,

[0120] b) Derive image feature data from the current image,

[0121] c) Compare the current image feature data derived from the current image (which can be interchangeably referred to as "current image feature data") with the second reference image feature data derived from the second reference image,

[0122] d) Determine whether the current image feature data matches the second reference image feature data.

[0123] In response to determining that the current image feature data does not match the second reference image feature data in step d), the method continues by returning to step a) to obtain the next image captured by camera 20 with the camera settings set to the second camera settings. Then, this next image can be used as the current image to repeat steps b) to d).

[0124] In response to determining that the current image feature data matches the second reference image feature data, the processing device 24 determines at S24 that a change of the scene 10 from the second illumination condition to the first illumination condition has been detected, and in response to detecting this change, switches the camera 20 to the first camera settings. The current image in which the change to the first illumination condition has been detected can be referred to as the "second image". Similar to the "first image", the "second image" can be the first image in time captured after the change from the second illumination condition to the first illumination condition, or corresponds to or is at least within a relatively short time window (e.g., having a length corresponding to a relatively short frame sequence (such as within 20 frames or less, 10 frames or less, or 5 frames or less)) after the change from the second illumination condition to the first illumination condition to reduce the latency in detecting the change to the first illumination condition.

[0125] Similar to the first reference image feature data, the second reference image feature data can be derived in advance from a second reference image (which can be downsampled) and stored in the memory 26 associated with obtaining this second reference image as the second reference image feature data 34d. In this case, the second reference image feature data 34d can be retrieved from the memory 26. In the case where the second reference image 34c (which can be downsampled, e.g., and not the second reference image feature data) has been stored in the memory 26, the processing device 24 can derive the second reference image data from the stored second reference image 34c before making the comparison.

[0126] Combined Figures 4 to 5 the previous discussion of the current image feature data, the first image feature data, and the reference image feature data is correspondingly applied to the current image feature data, the second image feature data, and the second reference image feature data discussed in connection with Figure 6 the flowchart. In particular, each of the above different sets of image feature data can include the same type of feature data. In addition, the comparison between the current image feature data / second image feature data and the second reference image feature data can be performed in a corresponding manner.

[0127] Although examples have been described above in which the lighting conditions vary between two steady-state lighting conditions, it is contemplated that even in such an environment, the lighting conditions may also change gradually and less significantly from the corresponding steady-state lighting conditions. The automatic exposure algorithm can reliably track such smaller gradual changes, for example, by incremental and / or gradual changes in one or more exposure-related parameters of camera 20. Accordingly, Figures 4 to 6 the flowchart method in can be used in combination with the automatic exposure algorithm to quickly and exactly adapt to a more sudden change between a first lighting condition and a second lighting condition as described above, and vice versa. For example, the processing device 24 can execute the automatic exposure algorithm in response to not detecting a change in the lighting conditions from the first lighting condition to the second lighting condition (or vice versa) to adapt one or more exposure-related parameters of camera 20 to a feasible smaller gradual change in the lighting conditions in scene 10. Thus, the method can continue to monitor scene 10 at S15 with camera 20 initially set to the second camera settings, and subsequently, use the automatic exposure algorithm to track the gradual change starting from the second lighting condition. Then, after the lighting conditions gradually return to the second lighting condition, a direct switch from the second camera settings to the first camera settings based on the second reference image can be used so that the camera settings return to the second camera settings (controlled by the automatic exposure algorithm), after which camera 20 can detect a sudden change from the second lighting condition to the first lighting condition based on the second reference image as described above. Accordingly, the method can continue to monitor scene 10 at S24 with camera 20 initially set to the first camera settings, and subsequently, use the automatic exposure algorithm to track the gradual change starting from the first lighting condition. Then, after the lighting conditions gradually return to the first lighting condition, a direct switch from the first camera settings to the second camera settings based on the first reference image can be used so that the camera settings return to the first camera settings (controlled by the automatic exposure algorithm), after which camera 20 can detect a sudden change from the first lighting condition to the second lighting condition based on the first reference image as described above.

[0128] Those skilled in the art will recognize that the present invention is in no way limited to the preferred embodiments described above. On the contrary, within the scope of the appended claims, many modifications and variations are possible. For example, the example scenarios discussed above are assumed to be monitored by an invariant or fixed camera. However, the methods described above can also be applied to cameras having a field of view adjustable by a pan-tilt device, where the method can further include sweeping the field of view of the camera across the environment between a set of fixed positions, where at each fixed position, the camera can monitor a different scene within the environment, and performing the control method aspects of the above method for each scene.

[0129] In addition, the methods described above can be extended to handle additional lighting conditions. For example, in addition to the first camera setting and the second camera setting, camera 20 can also determine a third camera setting suitable for monitoring scene 10 under a third lighting condition that is different from the first lighting condition and the second lighting condition. For example, the third lighting condition can correspond to a brightness level higher than both the first lighting condition and the second lighting condition, a brightness level lower than both the first lighting condition and the second lighting condition, or a brightness level between the first lighting condition and the second lighting condition. As a non-limiting example, in Figure 3 scene 12 of, the third lighting condition can correspond to door 2 being opened towards the brightly lit adjacent space 3, and artificial lighting 4 being turned on such that scene 12 is under an even brighter lighting condition.

[0130] The third camera setting can be determined by processing device 24 and stored in memory 26 for later retrieval. Similar to the first camera setting and the second camera setting, when scene 10 is under the third lighting condition, an automatic exposure algorithm can be used to determine the third camera setting. The third camera setting can include settings of the same exposure-related control parameters as the first camera setting and the second camera setting.

[0131] Camera 20 can further obtain a third reference image that represents scene 10 captured by camera 20 set to the first camera setting under the third lighting condition. The third reference image can generally be captured by camera 20 set to the first camera setting (more specifically, by image capture device 22 of camera 20) when scene 10 is under the third lighting condition (e.g., door 2 is opened, and Figure 3 the artificial lighting 4 in is turned on). That is, the third reference image can be captured by camera 20 using the first camera setting when scene 10 is under the third lighting condition. Therefore, when set to the first camera setting, the third reference image can accurately reflect scene 10 as seen or imaged by camera 20 under the third lighting condition.

[0132] Similar to the first camera setting and the second camera setting, as well as the first reference image and the second reference image, a third reference image can be obtained as part of the configuration or calibration steps of the method. That is, the scene 10 can be intentionally presented under a third lighting condition, the camera 20 can be set to the first camera setting (thus overriding the automatic exposure algorithm of the camera 20), and an image of the scene 10 can be captured by the camera 20. This image can be determined as the third reference image. Multiple images of the scene 10 can also be captured using the first camera setting, and the third reference image can be determined as the average image of these multiple images. Further contemplated, alternatively, the third reference image can be obtained as an image of the scene 10 captured by another but similar camera (e.g., by a camera installer when deploying a camera for monitoring the scene 10) using the corresponding camera setting and subsequently transmitted to the memory 26 of the camera 20.

[0133] Similar to the above discussion, the third reference image can optionally be downsampled, and / or the image feature data derived from the (downsampled) third reference image can be stored in the memory 26 as reference data 34 for subsequent use.

[0134] Accordingly, while monitoring the scene 10 with the camera 20 set to the first camera setting, the camera 20 (e.g., the processing device 24) can detect whether a change from the first lighting condition to the third lighting condition has occurred in the scene 10.

[0135] Similar to Figure 5 the method steps shown, a change from the first lighting condition to the third lighting condition can be detected. That is, the change from the first lighting condition to the third lighting condition can be detected by performing a comparison between: the first image feature data derived from the first image of the scene captured by the camera 20 set to the first camera setting after changing from the first lighting condition to the third lighting condition, and the third reference image feature data derived from the third reference image, to determine that the first image feature data matches the third reference image feature data.

[0136] In response to detecting this change (i.e., the match between the first image feature data and the third reference image feature data), the camera 20 can switch from the first camera setting to the third camera setting and continue to monitor the scene 10. Since the scene 10 is now under the third lighting condition, the captured image 28 of the scene 10 can be properly exposed.

[0137] The previous discussions regarding different types of image feature data, match scores, similarity thresholds, and ROIs associated with the first reference image feature data and the second reference image feature data are correspondingly applied to the third reference image feature data.

[0138] The third reference image can be supplemented with corresponding "shortcut" images in a similar manner to facilitate direct switching from one or more of the following: the third camera setting to the first camera setting (e.g., in response to detecting a change from the third lighting condition to the first lighting condition), the third camera setting to the second camera setting (e.g., in response to detecting a change from the third lighting condition to the second lighting condition), the second camera setting to the third camera setting (e.g., in response to detecting a change from the second lighting condition to the third lighting condition). It is further envisioned that the foregoing exemplary methods can be enhanced with one or more additional camera settings applicable to other lighting conditions (e.g., other steady-state lighting conditions) and corresponding reference / "shortcut" images to facilitate direct switching between them.

Claims

1. A method for controlling a camera for monitoring a scene, the method comprising: determining first camera settings for monitoring the scene under a first lighting condition and second camera settings for monitoring the scene under a second lighting condition different from the first lighting condition; obtaining a reference image, wherein the reference image represents the scene captured under the second lighting conditions with the camera set to the first camera settings; while monitoring the scene with the camera set to the first camera setting, detecting a change of the scene from the first lighting condition to the second lighting condition, wherein detecting the change from the first lighting condition to the second lighting condition comprises performing a comparison between: first image feature data derived from a first image of the scene captured with the camera set to the first camera setting after changing from the first lighting condition to the second lighting condition, and reference image feature data derived from said reference image, to determine a match between the first image feature data and the reference image feature data; as well as In response to detecting the change, switching the camera from the first camera setting to the second camera setting and continuing to monitor the scene.

2. The method according to claim 1, wherein: The first image characteristic data and the reference image characteristic data each include exposure related data.

3. The method according to claim 2, wherein: The exposure-related data comprises statistical data of pixel values ​​derived from the first image and the reference image, respectively, wherein the statistical data comprises one or more of: representative pixel values ​​such as average pixel values ​​or median pixel values ​​for the first image or the reference image, contrast, and frequency distribution of pixel values.

4. The method according to claim 1, wherein: The first image feature data and the reference image feature data include pixel values ​​of spatially corresponding pixels of the first image and the reference image.

5. The method according to claim 1, wherein: The first image feature data and the reference image feature data are derived from the first image and the reference image, respectively, by downsampling the corresponding image and extracting the first image feature data from the corresponding downsampled image.

6. The method according to claim 1, further comprising: The reference image characteristic data is derived and stored, wherein performing the comparison comprises comparing the first image characteristic data with the stored reference image characteristic data.

7. The method according to claim 1, wherein: The reference image is the scene captured under the second lighting conditions with the camera set to the first camera settings.

8. The method according to claim 1, wherein: The scene comprises one or more regions of interest, each region of interest being depicted in a respective one of the one or more portions in the first image and the reference image, and wherein the first image feature data and the reference image feature data are derived from at least the one or more portions in the first image and the reference image, respectively.

9. The method according to claim 8, wherein: The one or more regions of interest include objects such as doorways or artificial lighting having a first state and a second state, wherein the first state is associated with the first lighting condition and the second state is associated with the second lighting condition.

10. The method according to claim 1, wherein: The first lighting condition and the second lighting condition correspond to a first steady-state lighting condition and a second steady-state lighting condition having different brightness levels, respectively.

11. The method according to claim 1, wherein: The first camera setting and the second camera setting include settings of one or more exposure-related control parameters of the camera.

12. The method according to claim 1, wherein: The first camera setting and the second camera setting include settings of one or more of: shutter speed, aperture, ISO value, camera lighting, and IR filter status.

13. The method according to claim 1, further comprising: obtaining a second reference image, wherein the second reference image represents the scene captured under the first lighting conditions with the camera set to the second camera settings; while monitoring the scene with the camera set to the second camera setting, detecting a change of the scene from the second lighting condition to the first lighting condition, wherein detecting the change from the second lighting condition to the first lighting condition comprises performing a comparison between: second image feature data derived from a second image of the scene captured with the camera set to the second camera setting after changing from the second lighting condition to the first lighting condition, and second reference image feature data derived from the second reference image, to determine a match between the second image feature data and the second reference image feature data; as well as In response to detecting the change, switching the camera from the second camera setting to the first camera setting and continuing to monitor the scene.

14. A camera comprising: A processing device configured to execute the method according to claim 1.

15. A computer program product comprising computer program code portions configured, when executed by a processing device, to perform a method of controlling a camera for monitoring a scene, the method comprising: determining first camera settings for monitoring the scene under a first lighting condition and second camera settings for monitoring the scene under a second lighting condition different from the first lighting condition; obtaining a reference image, wherein the reference image represents the scene captured under the second lighting conditions with the camera set to the first camera settings; while the camera is set to the first camera setting and monitoring the scene, detecting a change of the scene from the first lighting condition to the second lighting condition, wherein detecting the change from the first lighting condition to the second lighting condition comprises performing a comparison between: first image feature data derived from a first image of the scene captured with the camera set to the first camera setting after changing from the first lighting condition to the second lighting condition, and reference image feature data derived from said reference image, to determine a match between the first image feature data and the reference image feature data; as well as In response to detecting the change, switching the camera from the first camera setting to the second camera setting and continuing to monitor the scene.

Images