阅读说明：本技术 一种基于云计算的车载图像特征自动识别方法与系统 (Vehicle-mounted image feature automatic identification method and system based on cloud computing ) 是由 陈泽通 于 2019-08-02 设计创作，主要内容包括：本发明提供一种基于云计算的车载图像特征自动识别方法与系统,通过云服务端对利用车载终端采集到的实时图像以及客户端上传的目标图像进行像素分解生成目标图像像素数据与关键帧像素数据,通过目标图像像素数据与关键帧像素数据的相似比与预设阈值的比较判断是否找到目标对象,进一步通过车载多媒体提醒用户,从而避免了用户在驾驶车辆时因为寻找目标对象而分心导致的安全隐患,实现了车辆的自主寻人与自动跟车提醒,还实现了客户端与车载终端的远程数据共享；通过人脸识别算法结合客户端与车载终端的远程数据共享,可实时监测当前车辆的周围环境,进行有效地远程车辆防盗。(The invention provides a vehicle-mounted image characteristic automatic identification method and system based on cloud computing, wherein a cloud service end carries out pixel decomposition on a real-time image acquired by a vehicle-mounted terminal and a target image uploaded by a client to generate target image pixel data and key frame pixel data, whether a target object is found is judged by comparing the similarity ratio of the target image pixel data and the key frame pixel data with a preset threshold value, and a user is further reminded through vehicle-mounted multimedia, so that the potential safety hazard caused by distraction of the user when the user searches for the target object during driving of a vehicle is avoided, the automatic person finding and automatic vehicle following reminding of the vehicle are realized, and the remote data sharing of the client and the vehicle-mounted terminal is also realized; the face recognition algorithm is combined with remote data sharing of the client and the vehicle-mounted terminal, so that the surrounding environment of the current vehicle can be monitored in real time, and remote vehicle theft prevention is effectively carried out.)

1. A vehicle-mounted image feature automatic identification method based on cloud computing is characterized by comprising the following steps:

s1, uploading the image captured by the vehicle-mounted camera to a cloud server in real time by the vehicle-mounted terminal;

s2, the cloud server receives the images and splices the images into a video segment at intervals of a preset time period and stores the video segment in a database;

s3, the cloud server side extracts key frames of each section of the video in sequence and carries out pixel decomposition on the key frames to obtain key frame pixel data;

s4, matching the key frame pixel data with target image pixel data provided by a client to obtain similarity;

and S5, the cloud server marks the key frames with the similarity reaching a preset threshold value, marks the key frames as effective information, stores the effective information in the database, and simultaneously issues the effective information to the vehicle-mounted terminal for display.

2. The automatic vehicle-mounted image feature recognition method based on cloud computing as claimed in claim 1, wherein before step S1, there is further provided the step of:

s01, the client uploads a target image to the cloud server;

s02, the cloud server side obtains the target image uploaded by the client side, and pixel decomposition is carried out on the target image to obtain target image pixel data.

3. The method for automatically identifying vehicle-mounted image features based on cloud computing as claimed in claim 2, wherein the step S1 specifically comprises the steps of:

s11, shooting images by the vehicle-mounted camera, wherein the images comprise images inside or outside a vehicle;

s12, the vehicle-mounted terminal compresses and codes the shot image and sends the image to the cloud server in real time through a streaming media technology and a wide area network;

and S13, the vehicle-mounted terminal caches the compressed and coded image in a local memory in a file form.

4. The cloud-computing-based automatic vehicle-mounted image feature recognition method according to claim 3, wherein in the step S4: the matching process of the target image pixel data and the key frame pixel data is based on a face recognition algorithm and a target tracking algorithm.

5. The method according to claim 4, wherein in the step S5, the cloud server issues the following information through a wide area network: the wide area network comprises a 4G network and a 5G network of FDD-LTE or TD-LTE.

6. The automatic vehicle-mounted image feature recognition method based on the cloud computing as claimed in claim 5, further comprising the steps of:

s6, the client accesses the cloud server, and views or downloads the video stored in the database or the key frames added with the effective information.

7. The vehicle-mounted image feature automatic identification system based on cloud computing is characterized by comprising a vehicle-mounted terminal, a cloud server and a client which are sequentially connected:

the client is used for uploading a target image to the cloud server;

the cloud server is used for acquiring a target image uploaded by the client and performing pixel decomposition on the target image to obtain target image pixel data;

the vehicle-mounted terminal is used for uploading the image captured by the vehicle-mounted terminal to the cloud server in real time;

the cloud server is also used for receiving the images and splicing the images into a section of video at intervals of a preset time period and storing the video into a database; extracting key frames of each section of the video in sequence, and performing pixel decomposition on the key frames to obtain key frame pixel data; matching the key frame pixel data with the target image pixel data to obtain similarity; and marking the key frames with the similarity reaching a preset threshold value as effective information, storing the effective information in the database, and simultaneously issuing the effective information to the vehicle-mounted terminal for displaying.

8. The cloud-computing-based automatic vehicle-mounted image feature recognition system according to claim 7, wherein: the vehicle-mounted terminal comprises an image acquisition module, an image processing module, a vehicle-mounted communication module and a cache module, wherein the image acquisition module, the image processing module and the vehicle-mounted communication module are sequentially connected, and the cache module is connected with the image processing module and the vehicle-mounted communication module;

the image acquisition module is used for transmitting the image captured by the vehicle-mounted camera to the image processing module in real time;

the image processing module is used for compressing and coding the shot image and then sending the image to the vehicle-mounted communication module through a streaming media technology;

the vehicle-mounted communication module is used for sending the image transmitted by the image processing module to the cloud server in real time through a wide area network; receiving the key frame marked as the effective information and issued by the cloud server;

the cache module is used for caching the compressed and encoded image in a local memory in a file form and storing the key frame marked as the effective information.

9. The cloud-computing-based automatic vehicle-mounted image feature recognition system according to claim 8, wherein: the vehicle-mounted terminal also comprises a multimedia module connected with the cache module and used for displaying the key frames marked as effective information.

10. The cloud-computing-based automatic vehicle-mounted image feature recognition system according to claim 9, wherein:

the image acquisition module is provided with a vehicle-mounted camera and is used for shooting the environment inside the vehicle and the environment outside the vehicle;

the wide area network comprises a 4G network and a 5G network of FDD-LTE or TD-LTE;

the matching process of the target image pixel data and the key frame pixel data is based on a face recognition algorithm and a target tracking algorithm;

the client is a mobile terminal used by a user.

Technical Field

The invention relates to the technical field of vehicle image processing, in particular to a vehicle-mounted image feature automatic identification method and system based on cloud computing.

Background

At present, after capturing an image, a vehicle-mounted camera performs decoding processing through a video decoding chip, performs data encoding (such as H.264 or MPEG) and caches the encoded image data in a local database, and sets a time threshold to generate a video file from the image data within a period of time and store the video file in the local database. However, the client accessing the local database for short-distance viewing, saving or playing the file needs to use a card reader or WiFi direct connection, which is very inconvenient to operate. In addition, the conventional vehicle-mounted terminal can only realize simple lane departure reminding, 360-degree panoramic looking around and the like through vehicle-mounted camera shooting, and does not recognize and screen images. The main disadvantages of current vehicle image processing are:

1. the local hardware chip only supports the transmission processing, coding and storage of images, has limited processing capability and low operation speed, and hardware equipment needs to be replaced if the algorithm needs to be expanded and the operation speed needs to be improved;

2. the current image storage is limited by the size of local storage space, when the space storage is insufficient, the required document must be manually saved or the earliest document must be automatically cleared, and hardware damage and document loss may be caused;

3. the stored image can be acquired only by contacting a connected vehicle-mounted central control or a near field communication protocol in a short distance, and the remote viewing cannot be realized.

Disclosure of Invention

The invention provides a vehicle-mounted image feature automatic identification method and system based on cloud computing, and solves the technical problems that an existing vehicle-mounted local hardware chip is weak in file processing capacity, a vehicle-mounted image processing technology cannot screen effective information, and a user cannot remotely check vehicle information.

In order to solve the technical problems, the invention provides a vehicle-mounted image feature automatic identification method based on cloud computing, which comprises the following steps:

s01, the client uploads a target image to the cloud server;

s02, the cloud server side obtains the target image uploaded by the client side, and pixel decomposition is carried out on the target image to obtain target image pixel data.

S1, uploading the image captured by the vehicle-mounted camera to a cloud server in real time by the vehicle-mounted terminal;

s2, the cloud server receives the images and splices the images into a video segment at intervals of a preset time period and stores the video segment in a database;

s3, the cloud server side extracts key frames of each section of the video in sequence and carries out pixel decomposition on the key frames to obtain key frame pixel data;

s4, matching the key frame pixel data with target image pixel data provided by a client to obtain similarity;

and S5, the cloud server marks the key frames with the similarity reaching a preset threshold value, marks the key frames as effective information, stores the effective information in the database, and simultaneously issues the effective information to the vehicle-mounted terminal for display.

S6, the client accesses the cloud server, and views or downloads the video stored in the database or the key frames added with the effective information.

The step S1 specifically includes the steps of:

s11, shooting images by the vehicle-mounted camera, wherein the images comprise images inside or outside a vehicle;

s12, the vehicle-mounted terminal compresses and codes the shot image and sends the image to the cloud server in real time through a streaming media technology and a wide area network;

and S13, the vehicle-mounted terminal caches the compressed and coded image in a local memory in a file form.

The matching process of the target image pixel data and the key frame pixel data is based on a face recognition algorithm and a target tracking algorithm.

The wide area network comprises a 4G network and a 5G network of FDD-LTE or TD-LTE.

Corresponding to the vehicle-mounted image characteristic automatic identification method based on cloud computing, the invention also provides a vehicle-mounted image characteristic automatic identification system based on cloud computing, which comprises a vehicle-mounted terminal, a cloud server and a client which are sequentially connected, wherein the vehicle-mounted terminal comprises a server and a server, and the server comprises the following components:

the client is used for uploading a target image to the cloud server;

the cloud server is used for acquiring the target image uploaded by the client and performing pixel decomposition on the target image to obtain target image pixel data.

The vehicle-mounted terminal is used for uploading the image captured by the vehicle-mounted terminal to the cloud server in real time;

the cloud server is also used for receiving the images and splicing the images into a section of video at intervals of a preset time period and storing the video into a database; extracting key frames of each section of the video in sequence, and performing pixel decomposition on the key frames to obtain key frame pixel data; matching the key frame pixel data with the target image pixel data to obtain similarity; and marking the key frames with the similarity reaching a preset threshold value as effective information, storing the effective information in the database, and simultaneously issuing the effective information to the vehicle-mounted terminal for displaying.

The vehicle-mounted terminal comprises an image acquisition module, an image processing module, a vehicle-mounted communication module and a cache module, wherein the image acquisition module, the image processing module and the vehicle-mounted communication module are sequentially connected, and the cache module is connected with the image processing module and the vehicle-mounted communication module;

the image acquisition module is used for transmitting the image captured by the vehicle-mounted camera to the image processing module in real time;

the image processing module is used for compressing and coding the shot image and then sending the image to the vehicle-mounted communication module through a streaming media technology;

the vehicle-mounted communication module is used for sending the image transmitted by the image processing module to the cloud server in real time through a wide area network; receiving the key frame marked as the effective information and issued by the cloud server;

the cache module is used for caching the compressed and encoded image in a local memory in a file form and storing the key frame marked as the effective information.

The vehicle-mounted terminal also comprises a multimedia module connected with the cache module and used for displaying the key frames marked as effective information.

The image acquisition module is provided with a vehicle-mounted camera and is used for shooting the environment inside the vehicle and the environment outside the vehicle;

the wide area network comprises a 4G network and a 5G network of FDD-LTE or TD-LTE;

the matching process of the target image pixel data and the key frame pixel data is based on a face recognition algorithm and a target tracking algorithm;

the client is a mobile terminal used by a user.

The invention provides a vehicle-mounted image characteristic automatic identification method and system based on cloud computing, wherein a cloud service end carries out pixel decomposition on a real-time image acquired by a vehicle-mounted terminal and a target image uploaded by a client to generate target image pixel data and key frame pixel data, whether a target object is found is judged by comparing the similarity ratio of the target image pixel data and the key frame pixel data with a preset threshold value, and a user is further reminded through vehicle-mounted multimedia, so that the potential safety hazard caused by distraction of the user when the user searches for the target object during driving of a vehicle is avoided, the automatic person finding and automatic vehicle following reminding of the vehicle are realized, and the remote data sharing of the client and the vehicle-mounted terminal is also realized; the face recognition algorithm is combined with remote data sharing of the client and the vehicle-mounted terminal, so that the surrounding environment of the current vehicle can be monitored in real time, and remote vehicle theft prevention is effectively carried out; by utilizing the strong background operation capability of the cloud server, effective information can be quickly and accurately screened from a large number of useless real-time images occupying a large amount of memory space and stored in a background database or sent to a vehicle-mounted terminal, so that the identification degree of the effective information is optimized, and the stability of file storage is improved.

Drawings

Fig. 1 is a flowchart illustrating a method for automatically identifying vehicle-mounted image features based on cloud computing according to an embodiment of the present invention;

fig. 2 is a flowchart of a target image pixel data acquisition work flow of a cloud computing-based vehicle-mounted image feature automatic identification method according to an embodiment of the present invention;

fig. 3 is a specific workflow of step S1 in fig. 1 according to an embodiment of the present invention;

fig. 4 is a system framework diagram of an automatic vehicle-mounted image feature identification system based on cloud computing according to an embodiment of the present invention.

The system comprises a vehicle-mounted terminal 1, an image acquisition module 11, an image processing module 12, a vehicle-mounted communication module 13, a cache module 14 and a multimedia module 15, wherein the vehicle-mounted communication module is used for communicating with the vehicle-mounted terminal; a cloud server 2; a client 3.

Detailed Description

The embodiments of the present invention will be described in detail below with reference to the accompanying drawings, which are given solely for the purpose of illustration and are not to be construed as limitations of the invention, including the drawings which are incorporated herein by reference and for illustration only and are not to be construed as limitations of the invention, since many variations thereof are possible without departing from the spirit and scope of the invention.

The vehicle-mounted image feature automatic identification method based on cloud computing, as shown in fig. 1 and 2, includes the following steps:

s01, the client 3 uploads a target image to the cloud server 2;

s02, the cloud server 2 obtains the target image uploaded by the client 3, and carries out pixel decomposition on the target image to obtain the target image pixel data.

S1, the vehicle-mounted terminal 1 uploads the image captured by the vehicle-mounted camera to the cloud server 2 in real time;

s2, the cloud server 2 receives the images and splices the images into a video segment at intervals of a preset time period and stores the video segment in a database;

s3, the cloud server 2 extracts key frames of each section of the video in sequence and carries out pixel decomposition on the key frames to obtain key frame pixel data;

s4, matching the key frame pixel data with target image pixel data provided by the client 3 to obtain similarity;

and S5, the cloud server 2 marks the key frames with the similarity reaching a preset threshold value, marks the key frames as effective information, stores the effective information in the database, and simultaneously issues the effective information to the vehicle-mounted terminal 1 for display.

S6, the client 3 accesses the cloud server 2, and views or downloads the video stored in the database or the key frame to which the valid information is added.

Referring to fig. 3, the step S1 specifically includes the steps of:

s11, shooting images by the vehicle-mounted camera, wherein the images comprise images inside or outside a vehicle;

s12, the vehicle-mounted terminal 1 compresses and encodes the shot image and sends the image to the cloud service terminal 2 in real time through a streaming media technology and a wide area network;

and S13, the vehicle-mounted terminal 1 caches the compressed and coded image in a local memory in a file form.

The matching process of the target image pixel data and the key frame pixel data is based on a face recognition algorithm and a target tracking algorithm.

The wide area network comprises a 4G network and a 5G network of FDD-LTE or TD-LTE.

Referring to fig. 4, in correspondence to the above vehicle-mounted image feature automatic identification method based on cloud computing, the present invention further provides a vehicle-mounted image feature automatic identification system based on cloud computing, including a vehicle-mounted terminal 1, a cloud server 2, and a client 3, which are connected in sequence:

the client 3 is configured to upload a target image to the cloud server 2;

the cloud server 2 is configured to obtain a target image uploaded by the client 3, and perform pixel decomposition on the target image to obtain target image pixel data.

The vehicle-mounted terminal 1 is used for uploading the image captured by the vehicle-mounted terminal to the cloud server 2 in real time;

the cloud server 2 is further configured to receive the images and splice the images into a section of video every preset time period, and store the video into a database; extracting key frames of each section of the video in sequence, and performing pixel decomposition on the key frames to obtain key frame pixel data; matching the key frame pixel data with the target image pixel data to obtain similarity; and marking the key frames with the similarity reaching a preset threshold value as effective information, storing the effective information in the database, and simultaneously issuing the effective information to the vehicle-mounted terminal 1 for display.

The vehicle-mounted terminal 1 comprises an image acquisition module 11, an image processing module 12, a vehicle-mounted communication module 13 and a cache module 14, wherein the image acquisition module, the image processing module 12 and the vehicle-mounted communication module 13 are sequentially connected, and the cache module 14 is connected with the image processing module 12 and the vehicle-mounted communication module 13;

the image acquisition module 11 is used for transmitting the image captured by the vehicle-mounted camera to the image processing module 12 in real time;

the image processing module 12 is configured to compress and encode the captured image and send the encoded image to the vehicle-mounted communication module 13 through a streaming media technology;

the vehicle-mounted communication module 13 is configured to send the image transmitted by the image processing module 12 to the cloud server 2 in real time through a wide area network; receiving the key frame marked as the effective information and sent by the cloud server 2;

the cache module 14 is configured to cache the compressed and encoded image in a local memory in a file form, and store the key frame marked as the valid information.

The vehicle-mounted terminal 1 further comprises a multimedia module 15 connected with the cache module 14 and used for displaying the key frames marked as valid information; the multimedia module 15 is further configured to perform human-computer interaction through a voice broadcasting module.

The image acquisition module 11 is provided with a vehicle-mounted camera and is used for shooting the environment inside the vehicle and the environment outside the vehicle; the wide area network comprises a 4G network and a 5G network of FDD-LTE or TD-LTE;

the matching process of the target image pixel data and the key frame pixel data is based on a face recognition algorithm and a target tracking algorithm;

the client 3 is a mobile terminal used by a user, and the mobile terminal comprises a mobile phone, a tablet and a computer.

The extraction method of the key frame comprises a sampling extraction method, a shot segmentation method and a cluster analysis method.

The specific workflow of the system is as follows:

firstly, the cloud service end 2 performs pixel decomposition according to a target image uploaded by the client end 3 through a wide area network to obtain target image pixel data, wherein the target image comprises a target person and a target vehicle.

Subsequently, the vehicle-mounted terminal 1 determines that the image acquisition module 11, i.e. the vehicle-mounted camera, acquires a real-time image inside or outside the vehicle as required, compresses and encodes the image into a file in a streaming media format through the image processing module 12, uploads the file to the cloud server 2 in real time through the wide area network and the streaming media technology through the vehicle-mounted communication module 13, and simultaneously stores the file in the cache module 14.

In the present embodiment, the image capturing module 11 is preferably a 360-degree panoramic automobile data recorder.

The cloud server receives the images uploaded from the vehicle-mounted terminal in real time, the images are spliced into a section of video at intervals of a preset time period and stored in a database, meanwhile, key frames in each second are sequentially intercepted according to an uploading sequence, pixel decomposition is carried out on the key frames to obtain key frame pixel data, image transmission in each second comprises 30 frames of pictures, and 2-3 frames of the pictures are extracted to serve as key frames. At this time, the cloud server 2 matches the key frames with the target image pixel data in sequence according to the target image pixel data serving as a comparison template to obtain a similarity ratio, compares the similarity ratio with a preset threshold, marks an area with a high similarity ratio with the target image on the key frame when the similarity ratio reaches the preset threshold, marks the key frame as effective information, stores the key frame in the database, simultaneously issues the key frame to the vehicle-mounted terminal 1, and performs display and voice prompt through the multimedia module 15. For example, if the target image is a target vehicle, the target vehicle is highlighted in the road image acquired by the vehicle-mounted camera in real time and displayed on the display interface of the multimedia module 15, so that a user can accurately capture the position of the front car-following, and the position of the front car-following can be described through the existing calibration technology, for example, "the target vehicle runs in a highway 50 m ahead"; if the target image is a target person, the panoramic image acquired by the vehicle-mounted camera is marked through a face recognition algorithm and displayed on a display interface of the vehicle-mounted multimedia module 15, so that a user can accurately capture the specific position of the target person, and meanwhile, route planning is performed again according to the position of the target person.

Meanwhile, the client 3 can view and download the cached video and the valid information stored in the cloud server 2 through the wide area network.

In addition, if the user needs to remotely prevent theft, the user can upload a preset scene image, wherein the preset scene comprises pictures of vehicle operation or part disassembly and stealing and the like which are not performed by the user, and the user can connect the cloud service terminal 2 through the 4G wide area network of the mobile phone client terminal 3 to check or download key frames similar to the preset scene image, so that the vehicle is prevented from being stolen; or through presetting an anti-theft mode, pre-storing a driver database, uploading the driver database to the cloud server 2 through the client, and when the vehicle is stolen, the cloud server 2 recognizes that the driver acquired by the image acquisition module 11 does not belong to a preset crowd in the driver database through a face recognition algorithm, actively sending the key frame with the similarity ratio lower than a preset threshold value to the user client for reminding, and simultaneously starting the automobile self-locking.

Meanwhile, a preset scene image of a vehicle porcelain-touching or vehicle accident can be uploaded, and scene key protection with key definition is carried out on related key frames when an event occurs.

The embodiment of the invention provides a vehicle-mounted image feature automatic identification method and system based on cloud computing, wherein a cloud server 2 is used for carrying out pixel decomposition on a real-time image acquired by a vehicle-mounted terminal 1 and a target image uploaded by a client 3 to generate target image pixel data and key frame pixel data, whether a target object is found is judged by comparing the similarity ratio of the target image pixel data and the key frame pixel data with a preset threshold value, and a user is further reminded through vehicle-mounted multimedia, so that the potential safety hazard caused by distraction when the user searches the target object when driving a vehicle is avoided, the autonomous person seeking and automatic vehicle following reminding of the vehicle are realized, and the remote data sharing of the client 3 and the vehicle-mounted terminal 1 is also realized; the face recognition algorithm is combined with remote data sharing of the client 3 and the vehicle-mounted terminal 1, so that the surrounding environment of the current vehicle can be monitored in real time, and remote vehicle theft prevention is effectively carried out; by utilizing the strong background operation capability of the cloud server 2, effective information can be quickly and accurately screened from a large number of useless real-time images occupying a large amount of memory space and stored in a background database or sent to the vehicle-mounted terminal 1, so that the identification degree of the effective information is optimized, and the stability of file storage is improved.

The above embodiments are preferred embodiments of the present invention, but the present invention is not limited to the above embodiments, and any other changes, modifications, substitutions, combinations, and simplifications which do not depart from the spirit and principle of the present invention should be construed as equivalents thereof, and all such changes, modifications, substitutions, combinations, and simplifications are intended to be included in the scope of the present invention.