WO2023109867A1

WO2023109867A1 - Camera image transmission method and apparatus, and camera

Info

Publication number: WO2023109867A1
Application number: PCT/CN2022/139015
Authority: WO
Inventors: 匡小冬; 刘琳; 白涛; 许俊峰; 闫冬升
Original assignee: 华为技术有限公司
Priority date: 2021-12-16
Filing date: 2022-12-14
Publication date: 2023-06-22

Abstract

Provided is a camera image transmission method, comprising: extracting a plurality of whole images from a video stream photographed by a camera, wherein the plurality of whole images record a plurality of targets; extracting a plurality of local images from the plurality of whole images, wherein there are a plurality of local images extracted from at least one whole image; and transmitting the extracted local images to a storage device, and transmitting, to the storage device, a whole image corresponding to the transmitted local images, wherein a plurality of whole images for local images correspond to a same whole image.

Description

Camera image transmission method device and camera

This application claims the priority of the Chinese patent application with the application number 202111540940.6 and the title of "camera image transmission method" filed with the State Intellectual Property Office of China on December 16, 2021, and claims to be filed in China on December 30, 2021 The State Intellectual Property Office, the priority of the Chinese patent application with the application number 202111645121.8 and the application name "camera image transmission method device and camera", the entire content of which is incorporated in this application by reference.

technical field

The invention relates to camera technology, in particular to a camera image sending method.

Background technique

The camera converts the optical signal into visualized images and videos through photoelectric conversion, and the visualized images and videos can be recorded on the storage medium for easy retrospective, so it is widely used in intersections, airports and other fields that require camera monitoring.

The network camera (IP camera, IPC) adds a network coding module and a network port on the basis of the traditional camera. The network encoding module compresses and encodes the collected data, and uses the network port to transmit through the network, and the remote storage server stores the encoded data after receiving it. It is used to view or process data on the storage server.

Take the capture network camera installed at the intersection as an example, it will frequently capture the vehicles passing through the shooting range of the network camera. The captured image is called a full image, and a partial image (such as a vehicle partial image) is extracted from the full image. If multiple images of the same vehicle are captured in the video stream captured by the same camera, the partial image with the highest recognition degree of the vehicle and the corresponding full image are selected for transmission, and the rest of the images are not transmitted.

For a busy intersection, after a long time of shooting, the network camera will generate a large number of partial and full images that need to be transmitted. On the urban road network, a large number of network cameras work in parallel at the same time, so a large number of images will be generated. Transmitting these images to the storage server undoubtedly poses a huge challenge to both the network transmission capability and the storage capacity of the storage server.

Therefore, how to minimize the number of images to be transmitted is an urgent problem to be solved at present.

Contents of the invention

In a first aspect, a camera image transmission method is provided, comprising: extracting multiple full images recorded with multiple targets from a video stream, wherein each full image records at least one target; obtaining a first group of partial images according to the multiple full images, The first group of partial images has multiple partial images, wherein different partial images record different targets, and at least two partial images in the first group of partial images are obtained from the same full image in the plurality of full images; Sending the first group of partial images and the full images corresponding to each partial image in the first group of partial images to a storage device, wherein the number of the sent full images is smaller than the number of the plurality of partial images.

Applying the solution provided in the first aspect, part of the local images jointly correspond to the same full image, so the number of full images sent by the camera to the storage device is less than the number of partial images. Therefore, the number of full images to be sent is reduced, and the occupation of network bandwidth and storage space of the storage device is reduced.

In a first possible implementation manner of the first aspect, the first group of partial maps includes a first partial map, and the first partial map is used to record a first target object, wherein the best image quality of the The first object is located in a second ensemble different from said first ensemble. It can be seen from this scheme that we do not use the best quality of the partial graph as the standard when we choose the partial graph. Optionally, the image quality of the first partial image meets a preset quality requirement.

In a second aspect, a camera image sending device is provided, including: a full image acquisition module, configured to extract multiple full images recorded with multiple targets from a video stream, wherein each full image records at least one target; a partial image acquisition module, It is used to obtain a first group of partial images based on the plurality of full images, the first group of partial images has multiple partial images, wherein different partial images record different targets, and at least two partial images in the first group of partial images The picture is obtained according to the same full picture in the plurality of full pictures; the sending module is configured to send the first group of partial pictures and the full picture corresponding to each partial picture in the first group of partial pictures to a storage device, wherein , the number of sent full images is less than the number of the plurality of partial images.

In the second aspect, a video camera is provided, including: a lens for receiving light; a sensor for photoelectric conversion to generate an original image; a processor system for processing the original image, and: how many records are extracted from the video stream A plurality of full images of a target, wherein each full image records at least one target; a first group of partial images is obtained according to the plurality of full images, and the first group of partial images has multiple partial images, wherein different partial images are recorded with different The goal is that at least two partial images in the first group of partial images are obtained from the same full image in the plurality of full images; the first group of partial images and each partial image in the first group of partial images correspond to The full images of are sent to the storage device, wherein the number of the sent full images is smaller than the number of the plurality of partial images.

In the fourth aspect, a program product is provided, which can be run in the processor of the camera. By running the program product, the camera can perform: extract multiple full pictures with multiple targets recorded from the video stream, wherein each full picture is recorded with at least A target; a first group of partial images is obtained according to the plurality of full images, the first group of partial images has multiple partial images, wherein different partial images record different targets, and at least two of the first group of partial images The partial image is obtained according to the same full image in the plurality of full images; the first group of partial images and the full images corresponding to each partial image in the first group of partial images are sent to the storage device, wherein the sent The number of full images is smaller than the number of the plurality of partial images.

Description of drawings

FIG. 1 is a structural diagram of an embodiment of components of a camera.

Fig. 2 is a flow chart of the camera sending images.

Fig. 3 is a schematic diagram of an embodiment of an image sent by a camera.

Fig. 4 is a flow chart of the camera sending images.

Fig. 5 is a schematic diagram of an embodiment of an image sent by a camera.

Fig. 6 is a schematic diagram of an embodiment where the camera is an image sending device.

Detailed ways

A video camera is an electronic device that can take pictures and record video. Referring to FIG. 1 , components constituting a camera according to an embodiment of the present invention include a lens 11 (lens), a sensor (sensor) 12 and a system on chip (SoC) 13 . The lens 11 is used to pass light; the sensor 12 is used to convert the optical signal from the lens into an electrical signal, and record it in the form of an original image; the system-on-chip 13 can encode the original image, such as converting RAW Images are encoded into JPEG images, HEIF images, H.264 videos, H.265 videos, etc. that are easily recognized by human eyes. The images involved in the embodiments of the present invention may be photos or frames in a video stream.

The function of the lens 11 is to present the light image of the observed object on the sensor of the camera. The lens combines various optical parts (mirrors, transmissive mirrors, prisms) of different shapes and media (plastic, glass or crystal) in a certain way, so that after the light is transmitted or reflected by these optical parts, according to people's It is necessary to change the transmission direction of the light to be received by the receiving device to complete the optical imaging process of the object. Generally speaking, the lens is composed of multiple groups of lenses with different curvatures at different pitches. The selection of spacing, lens curvature, light transmittance and other indicators determine the focal length of the lens. The main parameters of the lens include: effective focal length, aperture, field of view, distortion, relative illuminance, etc. The values of each index determine the overall performance of the lens.

The sensor 12 is a device that converts optical images into electronic signals, and is widely used in cameras and other electronic optical devices. Common sensors include: charge-coupled device (CCD) and complementary metal-oxide semiconductor (complementary MOS, CMOS). Both CCD and CMOS have a large number (for example, tens of millions) of photodiodes, and each photodiode is called a photosensitive unit, and each photosensitive unit corresponds to a pixel. During exposure, the photosensitive diode converts the light signal into an electrical signal including brightness (or brightness and color) after being irradiated by light, and the image is reconstructed accordingly. The Bayer array is a common image sensor technology. The Bayer array uses a Bayer filter to make different pixels sensitive to one of the three primary colors of red, blue, and green. These pixels are interleaved and then demosaiced (demosaicing) interpolation algorithm to get the original image. The Bayer array can be applied to CCD or CMOS, and the sensor using the Bayer array is also called a Bayer sensor.

In some cases, the same camera can be provided with multiple sensors, such as a dual-lens dual-sensor camera, a single-lens dual-sensor camera, and a single-lens triple-sensor camera. These sensors are used to image the same object to be photographed. When the number of lenses is less than the number of sensors, a beam splitter can also be set between the lens and the sensor, so that the light entering from one lens can be split to multiple sensors to ensure that each sensor can receive light.

The processor system is specifically a combination of multiple separate chips, or an integration of multiple chips (such as the system on chip 13). The image signal processor (ISP) 131 is used for performing image signal processing on the image, improving image contrast, saturation, gain and other image parameters for adjustment. The image signal processor 132 can fuse images, for example, fuse a frame of grayscale image and a frame of color image to generate a frame of high-quality color image to improve the image effect. The encoder 133 is used to encode images into images available for playback. For example, YUV data is generated into H.264 or H.265 frames. Images can be played on a display or stored. Multiple images can be played continuously to form a dynamic video. In some embodiments, these chips may not be integrated together, but are independent.

In addition, the camera may also have other components, such as optical filters, network interfaces, encoders, etc., which will not be described in detail here.

After the camera captures the image, it needs to send the full image and the partial image of the area where the target is located to the storage server. A solution will be described in more detail below in conjunction with FIG. 2 and flowchart 3 .

In step S11, the camera shoots a video stream, and the camera extracts a plurality of frames from the captured video stream.

The frames captured by the camera in sequence include: image 21 (full image), image 22 (full image) and image 23 (full image). The objects recorded in image 21 include: vehicle A, vehicle B, and vehicle C; the objects recorded in image 22 include: vehicle B and vehicle C, the vehicles recorded in image 23 include vehicle C, and no vehicle is recorded in image 24. Among them, vehicle B has the highest quality in image 22 , and vehicle C has the highest quality in image 23 .

Step S12 , the camera extracts partial images of the vehicle from the full image according to the image quality of the target, and obtains image 211 (partial image), image 212 (partial image), and image 231 (partial image).

Step S13 , the camera sends all the partial images and the full images where all the partial images are extracted to the storage server. The number of partial images in FIG. 1 is 3, and the corresponding number of full images is also 3. Therefore, a total of 6 images need to be sent to the storage server, and the 6 images are: image 211 , image 221 , image 231 , and image 21 , image 22 and image 23 . After the storage server receives the image sent by the camera, it can cache or persistently store the received image. Optionally, the storage server also has functions such as image optimization and image recognition.

It should be noted that, in this solution, multiple partial images come from the same full image, and each partial image still needs to be sent separately for the full image, so there is a case where the same full image is sent repeatedly.

In the above embodiment, a single camera needs to send a total of 6 images (3 full images + 3 partial images) to the storage server based on a short video stream. When multiple cameras are shooting for a long period of time, the total amount of images that need to be sent can be enormous. Therefore, the present invention further provides an embodiment in which the camera sends images. After the camera captures the image, when the same image appears in multiple vehicles, it can only send the partial images and a single full image of these vehicles instead of Send the full graph for each partial graph. It is equivalent to "sharing" the same full image with multiple partial images. As a result, the number of partial images and full images no longer maintains a 1:1 relationship, but the number of full images is less than the number of partial images, thereby reducing the number of full images sent.

The following process will specifically describe this embodiment with reference to FIG. 4 and FIG. 5 .

Step S21, the video camera shoots the video stream in real time, and saves the video stream in a local storage (internal memory, memory card, SSD or hard disk). The camera extracts an image (also called a frame) from the captured video stream. In order to distinguish it from a partial image, the complete image extracted in this step is called a full image. The extracted full images are shown in Fig. 21, Fig. 22 and Fig. 23 in descending order according to the shooting time.

Step S22, extracting the area where the target is located from the plurality of full images (big images) to generate a partial image (small image), and putting the partial image into a cache queue. The target is the target that needs to be observed or analyzed in the frame, such as moving objects such as cars, pedestrians, and animals.

The camera extracts an image from every two images in the video, obtains the multiple images, and detects the target in the extracted image (specifically, the area where the target is located in the image, for convenience) description, referred to here as the image quality of the target), for the same target appearing in different images, the following operations are performed:

(1) Cut out the partial image of the object with the highest quality, obtain the best quality partial image, and store it in the best cache queue. Therefore, when the target is photographed for the first time, the partial image is directly extracted from the frame and stored in the optimal cache queue. For a target that is not captured for the first time, if the quality of the target in the later captured frame is higher than the quality of the target in the earlier captured frame, the partial map of the target is extracted from the later captured frame to cover the best queue Existing local graphs in .

Each target can have an optimal cache queue for storing optimal local graphs. Put the full image (best full image) of the best quality partial image into the best cache queue, or put it into a separately set full image queue.

Put the frame number of the best full image (or other marks that can distinguish different full images) into the frame number queue.

(2) For the image in the large image, if the quality of the object is qualified, but if the object is not the highest quality object, the local image is cut out and stored in the suboptimal queue. Therefore, if the quality of the partial image extracted later is lower than that of the partial image extracted earlier, and the quality is qualified, the partial image extracted later will be put into the suboptimal queue.

The quality of the target can be judged from multiple dimensions, such as one of the following items: the pose of the target in the image, the image resolution of the area where the target is located, the display rate of key points of the target, or the recognition rate when recognizing the target, etc. When the target quality does not meet expectations, it will not be placed in the suboptimal cache queue.

According to the arrangement order of the images in the video stream, the images extracted from the video stream are image 21, image 22, image 23 and image 24 respectively. From these four images, it can be seen that as time goes by, vehicle A, vehicle B. The vehicle C drives out of the shooting range of the camera in sequence.

The image quality of each vehicle in different images is different, and the quality scores describing the quality are in brackets (the quality threshold is 50 points, and the task quality below 50 points does not meet expectations). The distribution is shown in Table 1 below, and the target and target scores described by the local graph are in parentheses.

Table 1

For image 21: If there is no record about vehicle A, vehicle B, and vehicle C in the optimal cache queue at this time, then directly extract image 211 (partial image), image 212 (partial image), image 212 (partial image), regardless of the quality of the target 213 (partial diagram), all three are put into the optimal cache queues corresponding to the vehicles respectively. Among them, each vehicle has a corresponding optimal cache queue and a second-best cache queue.

For image 22: image 221 is of higher quality than image 212, so image 221 is placed in the vehicle B optimal cache queue, covering image 212. The image 222 has a quality score of 40, so no extraction is required.

For image 23: cut out image 231 (partial image), because the quality of image 231 is higher than that of image 213, therefore put image 231 into the optimal cache queue of vehicle C, and cover image 213.

For image 24: Since no target (vehicle) is recorded, image extraction is not done.

Table 2 is a description of the images in the cache queue for each vehicle pair.

Table 2

Step S23, according to the partial images stored in the cache queue, select the partial images and provide the full images of the partial images to send to the storage device, and use the frame number queue to record the sent full images. This step can be executed after detecting that the vehicle disappears.

The strategy for selecting a partial image is: for any vehicle, if it is determined through the records in the frame number queue that a full image with this vehicle has been sent, and the quality of the vehicle in the previously sent full image meets the quality threshold ( The judgment quality threshold is an optional condition). Then, from the second-best queue, find the partial graph that satisfies the quality threshold and send it. For any vehicle mentioned above, no other full image can be transmitted additionally.

Based on the four images involved in Table 1, combined with Table 2, the images sent by the camera at different time points are described below.

(i) When the camera extracts image 22 from the video stream, it detects that vehicle A has disappeared relative to the previous image (image 21 is the image before image 22, and image 22 is the image after image 21). Therefore, the disappearance of vehicle A is taken as a trigger condition, and at this moment, an image related to vehicle A is selected for transmission.

It can be known that, for vehicle A, there is image 211 in the optimal cache queue, and the full image corresponding to image 211 is image 21 . According to the frame number queue of the whole image (the frame number queue is blank at this time), the camera has not sent the full image of the vehicle A before, so the image 211 and the image 21 are directly selected and sent to the storage device. And, record the frame number of the selected full picture (that is, the image 21) into the full picture frame number queue. In addition, the corresponding relationship between the image 211 and the image 21 is also sent to the storage device.

(ii) When the camera extracts image 23 from the video stream, it detects that vehicle B has disappeared relative to the previous image (image 22), so an image related to vehicle B needs to be sent.

For vehicle B, it is known from the queue of frame numbers of the whole image that the image 21 has been sent. In image 21 there is a partial view of vehicle B (image 212 ), and in image 21 the quality score of vehicle B is higher than 50 points. Although the quality score of vehicle B in image 22 is lower than that of vehicle B in image 21, for the consideration of "multiplexing" the whole image, image 212 is directly selected from the suboptimal cache and sent to the storage device (instead of Send the image 221 and the full image 22) where the image 221 is located, and send the correspondence between the image 212 and the image 21 to the storage device. That is to say, the two partial images (image 212 and image 211 ) correspond to the same full image (image 21 ).

Compared with the process described in steps S11-S13, there is such a difference: because the quality of vehicle B in FIG. 22 is higher than that of vehicle B in FIG. 21 . Therefore, what needs to be sent to the vehicle B is the best partial image 221 and the image 22 to which the best partial image 221 belongs. It can be seen that, compared with the embodiment described in steps S11-S13, this embodiment sends one less full image (image 22). The network load and the space occupation of the storage device are reduced.

(iii) When the camera captures the image 24 , it detects that the vehicle C disappears relative to the previous image (image 23 ), thus triggering the transmission of an image related to the vehicle C.

For vehicle C, it is known from the queue of frame numbers of the whole image that the image 21 has been sent. In image 21 there is a partial view of vehicle C (image 212 ), however in image 21 the quality score of vehicle C is below 50 points. However, the quality score of the image 231 is higher than the threshold, so for the vehicle C, the selected image 231 and the full image 23 corresponding to the image 231 are sent to the storage device. In addition, the corresponding relationship between the image 231 and the image 23 is also sent to the storage device.

It can be seen that, in this embodiment, the camera sends a total of 5 images to the storage device (3 partial images + 2 full images).

It should be noted that, for vehicle C, another optional situation is that the image quality is not considered. It is known from the frame number queue of the whole image that the image 21 has been sent, so the image 213 is directly sent (and the association between the image 213 and the image 21 is sent), and the image 231 and the image 23 are not selected for sending. In this case, the camera sends a total of 4 full images to the storage device (3 partial images + 1 full image).

In actual use, for a specific vehicle, step 22 is executed first, and then step 23 is executed. Considering that the camera will continuously extract a large number of images from the video stream, the operation of extracting images and sending images is actually performed continuously.

In this embodiment, the partial images sent to the storage device may be referred to as the first group of partial images, and the partial images in the cache queue may be referred to as the second group of partial images. In the second group of partial graphs, some partial graphs record the same target, and in the second group of partial graphs, different partial graphs record different targets. Thus the first set of partial maps is part of the second set of partial maps, the number of partial maps in the first set of partial maps being less than the number of partial maps in the second set of partial maps.

The present invention also provides a camera image sending device, which can execute the above method. The device is, for example, a camera, or a software module in the camera, and the software module can be run by an image signal processor (ISP) 131 .

The image sending device includes a full image acquisition module 31 , a partial image acquisition module 32 and a sending module 33 .

The full image acquisition module 31 may execute step S21, and the partial image acquisition module 32 is configured to extract multiple full images recorded with multiple objects from the video stream, wherein each full image records at least one object. The partial image acquisition module 32 can perform the process of selecting partial images and full images in step S22 and step S23, the partial image acquisition module 32 is used to obtain a first group of partial images according to the multiple full images, and the first group of partial images has multiple Partial images, wherein different partial images record different targets, and at least two partial images in the first group of partial images are obtained from the same full image among the plurality of full images. The sending module 33 may execute the sending process in step S23, for sending the first group of partial images and the full images corresponding to each partial image in the first group of partial images to the storage device, wherein the sent full The number of graphs is smaller than the number of the plurality of partial graphs.

The sending module 33 is further configured to, in the image sent to the storage device, establish a corresponding relationship between a partial image in the first group of partial images and a full image having the partial image.

The partial image acquisition module 32 can specifically be used to: obtain a second group of partial images according to the plurality of full images, and some partial images in the second group of partial images record the same target, wherein the partial images recording the same target come from Different full maps; selecting the first group of partial maps according to the second group of partial maps, where different partial maps record different targets in the first group of partial maps.

The present invention also provides a program product that can run in the video camera and is used to execute the aforementioned method steps S21-S23. One or more of the above modules or units may be realized by software, hardware or a combination of both. When any of the above modules or units is implemented by software, the software exists in the form of computer program instructions and is stored in the memory, and the processor can be used to execute the program instructions and realize the above method flow. The processor may include but not limited to at least one of the following: a central processing unit (central processing unit, CPU), a microprocessor, a digital signal processor (DSP), a microcontroller (microcontroller unit, MCU), or artificial intelligence Various types of computing devices that run software such as processors, each computing device may include one or more cores for executing software instructions to perform calculations or processing. The processor can be built into SoC (system on chip) or application specific integrated circuit (ASIC), or it can be an independent semiconductor chip. The core of the processor is used to execute software instructions for calculation or processing, and can further include necessary hardware accelerators, such as field programmable gate array (field programmable gate array, FPGA), PLD (programmable logic device) , or a logic circuit that implements a dedicated logic operation.

Referring to FIG. 1 , the present invention also provides an embodiment of a camera, including: a lens 11 for receiving light; a sensor 12 for photoelectric conversion to generate an original image; and a processor system 13 .

The processor system 13 is used to process the original image and execute the above-mentioned method embodiments. The method of the processor system 13 includes, for example: extracting a plurality of full images recorded with multiple targets from the video stream, wherein each full image records at least one target; obtaining a first group of partial images according to the plurality of full images, and the first group Partial graphs Multiple partial graphs, wherein different partial graphs record different targets, and at least two partial graphs in the first group of partial graphs are obtained from the same full graph in the multiple full graphs; The group of partial images and the full images corresponding to each partial image in the first group of partial images are sent to the storage device, wherein the number of the sent full images is smaller than the number of the plurality of partial images.

When the above modules or units are implemented in hardware, the hardware can be CPU, microprocessor, DSP, MCU, artificial intelligence processor, ASIC, SoC, FPGA, PLD, dedicated digital circuit, hardware accelerator or non-integrated discrete device Any one or any combination of them, which can run necessary software or not depend on software to execute the above method flow.

Claims

A camera image sending method, comprising:

extracting multiple full images recorded with multiple targets from the video stream, wherein each full image records at least one target;

Obtain a first group of partial images according to the multiple full images, the first group of partial images includes multiple partial images, wherein different partial images record different targets, and at least two partial images in the first group of partial images The figure is obtained from the same full picture of said multiple full pictures;

Sending the first group of partial images and the full images corresponding to each partial image in the first group of partial images to a storage device, wherein the number of the sent full images is smaller than the number of the plurality of partial images.
The method according to claim 1, characterized in that:

The first group of partial images includes a first partial image and a second partial image, and the sent full image includes a first full image, wherein the first partial image and the second partial image are derived from the first full image picture.
The method according to claim 1, wherein obtaining the first group of partial images according to the plurality of full images further comprises:

The first set of partial maps is determined based on image quality.
The method of claim 1, wherein the image quality is related to at least one of the following:

The pose of the target in the image;

Image Resolution;

Display rate of target key points in the image;

The target recognition rate in the image.
The method according to claim 1, wherein sending the first group of partial images specifically comprises:

The plurality of extracted full images are detected in time sequence, and for each target, when it is detected that a target disappears, it is triggered to obtain a partial image of the disappeared target and send it to the storage device.
The method according to claim 1, characterized in that:

In the images sent to the storage device, a correspondence relationship between the partial images in the first group of partial images and the full image owning the partial images is established.
The method according to claim 1, wherein said obtaining the first group of partial images according to said multiple full images comprises:

Obtaining a second group of partial images according to the plurality of full images, wherein some partial images in the second group of partial images record the same target, wherein the partial images recording the same target come from different full images;

The first group of partial maps is selected according to the second group of partial maps, and in the first group of partial maps, different partial maps record different targets.
A camera image sending device is characterized in that it comprises:

A full image acquisition module, configured to extract multiple full images recorded with multiple targets from the video stream, wherein each full image records at least one target;

A partial image acquisition module, configured to obtain a first group of partial images according to the plurality of full images, wherein the first group of partial images has multiple partial images, wherein different partial images record different targets, and the first group of partial images At least two of the partial images are obtained from the same full image in the plurality of full images;

A sending module, configured to send the first group of partial images and the full images corresponding to each partial image in the first group of partial images to a storage device, wherein the number of sent full images is smaller than the number of partial images in the first group The number of graphs.
The camera image sending device according to claim 8, characterized in that:

The first group of partial images includes a first partial image and a second partial image, the full image sent by the sending module includes the first full image, and both the first partial image and the second partial image are consistent with the first partial image A full graph corresponds, wherein the first partial graph and the second partial graph are derived from the first full graph.
The camera image sending device according to claim 8, wherein the partial image acquisition module is used for:

The first set of partial maps is determined based on image quality.
The camera image sending device according to claim 9 or 10, characterized in that in the first group of partial images, the image quality of some partial images is not the best image quality of the target in the plurality of full images.
The camera image sending device according to claim 8, wherein the partial image acquisition module is also used for:

In the images sent to the storage device, a correspondence relationship between the partial images in the first group of partial images and the full image owning the partial images is established.
The camera image sending device according to claim 8, wherein the partial image acquisition module is specifically used for:

Obtaining a second group of partial images according to the plurality of full images, wherein some partial images in the second group of partial images record the same target, wherein the partial images recording the same target come from different full images;

The first group of partial maps is selected according to the second group of partial maps, and in the first group of partial maps, different partial maps record different objects.
A camera comprising:

lens for receiving light;

Sensors for photoelectric conversion to generate raw images;

processor system for processing raw images, and:

extracting multiple full images recorded with multiple targets from the video stream, wherein each full image records at least one target;

A first group of partial images is obtained according to the plurality of full images, the first group of partial images has multiple partial images, wherein different partial images record different targets, and at least two partial images in the first group of partial images are Obtained from the same full image of the plurality of full images;

Sending the first group of partial images and the full images corresponding to each partial image in the first group of partial images to a storage device, wherein the number of the sent full images is smaller than the number of the plurality of partial images.