100 to be received in the second camera 100 may then either be prepared in a centralized solution using a central server or in a distributed solution using the first device and second camera 100. In the centralized solution, only the central server needs to know the location of the first device and the location and field of view of the second camera 100. After receiving information from the first device indicating location and direction of movement of the object, the central server may then prepare the information indicating that the object is approaching the field of view of the second camera 100 and provide this information to the second camera 100. In the distributed solution, the first device may send the information indicating location and direction of movement of the object directly to the second camera 100 and the second camera 100 may then, being aware of its own location and field of view, determine whether the object is approaching the field of view of the second camera 100, and when it is, trigger one or more of the three actions. In the latter case, the information indicating that the object is approaching the field of view of the second camera 100 would be the information indicating a location and a direction of movement of the object received from the first device.

In addition to indicating that the object is approaching the field of view of the second camera 100, the received information may further indicate a time when the object is estimated to appear in the field of view of the second camera 100 and/or a velocity of the object. By knowing a time when the object is estimated to appear in the field of view of the second camera 100, a better timing of the performing of one of the three actions prior to the object appearing in the field of view of the second camera 100 can be achieved. The first device may calculate the time when the object is estimated to appear in the field of view of the second camera 100 based on the velocity of the object and the location and field of view of the second camera.

Furthermore, the field of view of the second camera 100 may be logically divided into a set of subareas, wherein each subarea corresponds to a respective subarea of each image frame of the second camera 100. The received information in the second camera may then further indicate the subarea of the field of view of the second camera 100 in which the object is expected to first appear. By knowing a subarea, the one or more actions may be performed in relation to the subarea.

To determine the location and direction of movement of an object would typically require the first device to provide at least 2-dimensional representations over time. Hence, the first device may be one of, but not limited to, a first camera, a radar, a lidar, a thermal camera, and a sonar sensor. In such scenarios, in a centralized solution, the first device may determine a relative location and relative direction of movement in relation to its location and field of view by detection of the object over a sequence of image frames. The centralized server could then, knowing the absolute or relative location and field of view of both the first device and the second camera 100, determine whether the object is approaching the field of view of the second camera 100, and when it is, provide such information to the second camera 100. In a distributed solution, the knowledge of the location and field of view of the first device and second device in relation to a common coordinate system may be located in the respective devices. In alternative, the each device may have the relative location of the other device. The location and direction of movement of the object may then be determined in the first device by detection of the object over a sequence of image frames and information indicating the location and direction of movement of the object may be provided to the second camera 100. The second camera 100 may then, being aware of its own location and field of view, determine whether the object is approaching the field of view of the second camera 100, and when it is, trigger one or more of the three actions. In the latter case, the information indicating that the object is approaching the field of view of the second camera 100 would be the information indicating a location and a direction of movement of the object received from the first device. From the detections of the object over a sequence of image frames of the first device, an estimated velocity of the person (object) and/or an estimated time when the person (object) will enter the field of view of the second camera 100 can be determined. Furthermore, from the location and direction of movement, the subarea of the field of view of the second camera 100 in which the object will appear can be determined.

In some scenarios, the first device may include one or more sensors indicating only that an object has passed a specific point or line. For example, if the first device includes a sensor that can sense that a person has passed through a door, or passed a line, which leads into a corridor or other confined space, where the second camera 100 has a location and field of view at the other end of the corridor or other confined space, information from the first device that a person has passed the door or the line will then constitute information that the person (object) is approaching the field of view of the second camera 100 based on the knowledge of the location of the first device and the location and field of view of the second camera 100. If the first device includes one or more further sensors at known locations along the corridor or other confined space, an estimated velocity of the person (object) and/or a better estimation of the time when the person (object) will enter the field of view of the second camera 100 can be determined. In such scenarios, the field of view of the second camera 100 can be divided into subareas where one subarea is closest to the one or more sensors of the first device. In such a case, the subarea of the field of view of the second camera 100 in which the object will appear can be determined as the subarea closes to the one or more sensors of the first device.

Generally, it is not essential how the location of the first device and the location and field of view of the second device is determined as long as it is determined and then known either centrally or distributed such that the second camera 100 can receive information indicating that the object is approaching its field of view, wherein the information is based on information from the first device indicating the location and direction of movement of the object. For example, the location of the first device and the location and field of view of the second device may be determined at set up of the camera system, e.g., using gps positioning and/or location of mounting based on a map. In alternative the location of the first device and the location and field of view of the second device may be determined using a self-learning system which receives information of detection and movement of objects in the first device and the second camera and over time determines the location of the first device and the location and field of view of the second device and/or whether or not an object having a particular location and direction of movement according to the first device is approaching the field of view of the second camera 100.

One of the actions that may be triggered in response to the received information is generation of an intra frame or initiation of an intra refresh procedure. When the received information indicates that the object is approaching the field of view of the second camera 100, generation of an I-frame is triggered before the object appears in the field of view of the second camera 100. After such an I-frame has been generated, the quality of the background is enhanced which means that, when the object appears in the field of view of the second camera 100, there will be less need for I-blocks in relation to areas including only the background and not the object. An advantage of generating the I-frame before the object enters the field of view of the second camera 100 is that such an I-frame will typically be larger in number of bits in relation to a P-frame or B-frame. When the object then appears in the field of view of the second camera 100, there will be a need for generating I-blocks in relation to the object. Such I-blocks will increase the number of bits of the encoded image frame. Similarly, the movement of the object will also generally increase the number of bits of each encoded frame during the movement of the object in the field of view. Hence, by triggering the generation of an I-frame to update the background occur before the object appears in the field of view of the second camera 100, the increased number of bits due to this will be in different encoded frames than encoded frames including increased number of bits due to generation of I-blocks when the object appears and due to the movement of the object.

If the received information indicates the time when the object is estimated to appear in the field of view of the second camera 100, the generation of the I-frame or the intra refresh can be timed in relation to that time, namely before the object is estimated to appear in the field of vie of the second camera 100, preferably just before.

If the received information further indicates the subarea of the field of view of the second camera 100 in which the object is expected to first appear, generation of I-blocks can be adapted by means of an intra refresh procedure adapted to this.

Generally, an intra refresh procedure is a procedure where each image frame of a set of image frames is divided into a plurality of subareas and where one or more subareas of the plurality of subareas are intra encoded in a first image frame, different one or more subareas of the set of subareas are intra encoded in a second image frame, and so forth until all subareas of the plurality of subareas have been intra encoded over the set of image frames. Each subarea of the image frames corresponds to a respective one of subareas of the field of view of the second camera 100. As an example, a division of an image frame 300 into four subareas 310, 320, 330, 340 is shown in FIG. 3, wherein a first subarea 310 is an upper left portion, a second subarea 320 is an upper right portion, a third subarea 330 is a lower left portion, and a fourth subarea 340 is an lower right portion of the image frame 300. Other divisions are of course possible, such as a division into more subareas and other patterns, such as horizontal or vertical stripes or other.

Generating an intra frame or initiating an intra refresh procedure may then comprise first intra refresh, i.e., intra encoding, of subareas of image frames of the second camera 100 that do not correspond to the subarea of the field of view of the second camera 100 in which the object is estimated to first appear. This may be performed by intra encoding all subareas other than the subarea corresponding to the subarea of the field of view of the second camera 100 in which the object is expected to first appear in one image frame. In alternative this may be spread out over a number of image frames, e.g., such that each subarea of the subareas other than the subarea corresponding to the subarea of the field of view of the second camera 100 is intra encoded in a separate image frame all appearing before the object is expected to enter the field of view of the second camera 100. When the object has appeared in the field of view of the second camera 100, intra encoding is performed for the subarea of the image frame of the second camera 100 that corresponds to the subarea of the field of view of the second camera 100 in which the object is estimated to first appear. In relation to FIG. 3, if the object is expected to appear first in subarea 310, intra encoding is performed first for subareas 320, 330, 340 before the object appears in the image frame 300 and then, upon or subsequent to the object appearing in the image frame 300, intra encoding is performed for subarea 310. The intra encoding for the subareas 320, 330, 340 may occur in a single image frame or in separate image frames. In an alternative embodiment, the intra encoding of the subareas 320, 330, 340 is performed in a single or separate image frames is initiated at a time such that it is completed at a time such that the intra encoding of the subarea 310 can be performed before the object is estimated to appear in the subarea 310. Once the object appears in the subarea 310, at least a subset of blocks of the subarea 310 are encoded into I-blocks.

By triggering intra refresh in subareas of image frames of the second camera 100 that do not correspond to the subarea of the field of view of the second camera 100 in which the object is estimated to first appear before the object appears the field of view of the second camera 100, it is ensured that the background is up to date, and the quality of the background is enhanced in the subareas in which the object will not appear first. In this way, when the object appears in the field of view of the second camera 100, there will be less need for I-blocks in subareas of the field of view in which the object will not appear first. When the object appears in the field of view of the second camera 100, intra refresh is triggered in the subarea of the image frame of the second camera 100 in which the object appears. This will typically increase the number of bits of the encoded image frame. Similarly, movement of the object in relation to an image frame will also generally increase the number of bits of that image frame as encoded. Hence, the increased number of bits required due to intra refresh of the subareas other than the subarea in which the object will appear will be in different encoded image frames than the encoded image frame including an increased number of bits due intra refresh of the subarea corresponding to the subarea in which the object first appears and the encoded image frames including an increased number of bits due to movement of the object. Furthermore, since the subarea of an image frame will have to be updated with I-blocks when the object appears in the corresponding subarea of the field of view, the intra refresh of that subarea can be postponed until after the object has appeared. In some embodiments however, the intra refresh of that subarea is completed before the object appears, preferably as close as possible before the object appears.

If the received information further indicates the time when the object is estimated to appear in the field of view of the second camera 100, the intra encoding of the subareas can be timed in relation to that time, e.g., such that the subarea of an image frame corresponding to the subarea in which the object appears is intra encoded when, or as soon as possible after, the object appears in the field of view of the second camera 100. As indicated hereinabove, by an object appearing is meant that any portion of the object appears in the field of view of the second camera 100.

Another of the actions that may be triggered is adjusting one or more image processing parameters. Herein, image processing parameters are parameters for processing of an image frame once it has been captured in the camera. Examples of such processing are color balancing, white balancing, low light noise filtering, temporal noise filtering, and background segmentation.