阅读说明：本技术 一种用区块链提升跨摄像头行为识别可信度的方法 (Method for improving recognition reliability of cross-camera behaviors by using block chain ) 是由 薛素金 方勇 杨焜 于 2021-08-02 设计创作，主要内容包括：本发明公开了一种用区块链提升跨摄像头行为识别可信度的方法,包括摄像头模块、区块链模块和行为识别模块,包括以上步骤,步骤一：定义一个单链的区块链服务；步骤二：定义能够自动分割视频,计算哈希作为文件名存储,并能够将文件哈希打包交易后签名发送到区块链存证的摄像头物联网系统；步骤三：摄像头服务定期轮询区块高度,按照区块链每个块的出块时间来分割视频,区块间通过哈希相联系,单条链的区块中的数据按照区块高度的顺序递增,索引存证的视频按照出块顺序递增,同时可根据区块时间戳来确定事件发生的时间,确保事件先后顺序不会被打乱,本发明可提高跨摄像头行为识别的结果的可信度,降低行为识别结果被质疑的可能。(The invention discloses a method for improving cross-camera behavior identification reliability by using a block chain, which comprises a camera module, a block chain module and a behavior identification module, and comprises the following steps: defining a single-chain blockchain service; step two: defining a camera Internet of things system which can automatically divide a video, calculate Hash to be stored as a file name, and send a signature to a block chain certificate after file Hash is packaged and traded; step three: the camera service periodically polls the block height, divides videos according to the block outlet time of each block of the block chain, the blocks are connected through Hash, data in the blocks of the single chain are increased progressively according to the block height sequence, the videos for indexing and storing certificates are increased progressively according to the block outlet sequence, and meanwhile the time of occurrence of events can be determined according to the block timestamps, so that the sequence of the events can be ensured not to be disordered.)

1. A method for improving the identification reliability of cross-camera behaviors by using a block chain comprises a camera module, a block chain module and a behavior identification module, and is characterized by comprising the following steps:

the method comprises the following steps: defining a single-chain blockchain service;

step two: defining a camera Internet of things system which can automatically divide a video, calculate Hash to be stored as a file name, and send a signature to a block chain certificate after file Hash is packaged and traded;

step three: the camera service periodically polls the tile height and partitions the video according to the out-of-tile time of each tile in the chain of tiles.

2. The method of claim 1, wherein the method for improving confidence level of behavior recognition across cameras by using a block chain comprises: the blocks are connected through Hash, data in the blocks of the single chain are increased progressively according to the block height sequence, the video for indexing and storing the certificate is increased progressively according to the block shooting sequence, meanwhile, the occurrence time of the event can be determined according to the block timestamp, and the sequence of the event is guaranteed not to be disturbed.

3. The method of claim 2, wherein the method for improving confidence level of behavior recognition across cameras by using a block chain comprises: the uniqueness and the reliability of the video are determined through the Hash operation of the camera, and the video is determined to be strictly increased in time sequence through the block chain.

4. The method of claim 3, wherein the method for improving confidence level of behavior recognition across cameras by using a block chain comprises: the hash algorithm is an algorithm for mapping an input value to an output value (hash value) with a certain length, the hash value has the characteristics of high randomness and collision resistance, the hash value may correspond to different input values, but it is impossible to find different input values within a normal operation time, and it is impossible to find two inputs having the same hash value through a large number of operations.

5. The method of claim 4, wherein the method for improving confidence level of behavior recognition across cameras by using a block chain comprises: the hash can be regarded as an anti-counterfeit label of the input value, and the hash functions as the anti-counterfeit labels of different videos.

6. The method of claim 5, wherein the method for improving confidence level of behavior recognition across cameras by using a block chain comprises: the camera module comprises a camera, and the camera collects and segments videos to obtain video data.

7. The method of claim 6, wherein the method for improving confidence level of behavior recognition across cameras by using a block chain comprises: the block chain module comprises a block chain, the video data is subjected to camera calculation hash through the block chain, and is packaged for transaction chain linking, and then video HashCuve evidence on the chain is carried out.

8. The method of claim 7, wherein the method for improving confidence level of behavior recognition across cameras by using a block chain comprises: the behavior recognition module comprises a behavior recognition service, the behavior recognition service requests evidence storage information and acquires video data, the behavior recognition service performs cross-camera behavior analysis, the behavior recognition service links the used video hash and recognition results with evidence storage, and then outputs result data.

Technical Field

The invention relates to the field of behavior recognition credibility, in particular to a method for improving cross-camera behavior recognition credibility by using a block chain.

Background

Machine learning is machine learning as opposed to algorithm learning. The earliest application of machine learning is the resolution of junk mails, and the traditional idea for solving the problem of computers is to write rules, define ' junk mails ' and enable the computers to execute the ' junk mails, so that an algorithm needs to be written. The algorithm has fixed input and output, the input is that a mail is corresponding to all information, and the output is that whether the mail is a junk mail or not is judged.

At present, algorithms for recognizing behaviors of people through machine learning tend to be mature, but the machine learning is to perform corresponding calculation output on input of pictures or videos, and cannot judge the sequence of events, so that the accuracy of a machine learning result is difficult to ensure for behavior recognition with event sequence judgment, and therefore the invention provides a method for improving the reliability of cross-camera behavior recognition by using a block chain to solve the problems.

For example, application No. 201811353903.2 discloses an active video identification method in combination with a blockchain, which is applied to the field of video identification and prevents video frames from being tampered in order to store the video frames completely and truly; the invention combines the public block chain and the private block chain to encrypt and store the video file and verify the source and tampering of the video, and the private block chain can be used as a local database to provide a basis for verifying the source and tampering of the video; and the first transaction timestamp generated by the private blockchain can be used as the occurrence time of the abnormal event to ensure the immediacy of the time for locally storing the video content, but the invention cannot accurately judge the sequence of the occurrence of the event.

Disclosure of Invention

The invention aims to provide a method for improving the recognition reliability of cross-camera behaviors by using a block chain, so as to solve the technical problem.

In order to solve the technical problems, the invention adopts the following technical scheme:

a method for improving cross-camera behavior recognition reliability by using a block chain comprises a camera module, a block chain module and a behavior recognition module, and comprises the following steps:

the method comprises the following steps: defining a single-chain blockchain service;

step two: defining a camera Internet of things system which can automatically divide a video, calculate Hash to be stored as a file name, and send a signature to a block chain certificate after file Hash is packaged and traded;

step three: the camera service periodically polls the tile height and partitions the video according to the out-of-tile time of each tile in the chain of tiles.

Preferably, the blocks are connected through Hash, data in the blocks of a single chain are increased progressively according to the block height sequence, the video for indexing and storing the certificate is increased progressively according to the block shooting sequence, meanwhile, the occurrence time of the event can be determined according to the block timestamp, and the sequence of the event is guaranteed not to be disturbed.

Preferably, the uniqueness and reliability of the video are determined through hash operation of the camera, and the video is determined to be strictly increased in time sequence through the block chain.

Preferably, the hash algorithm is an algorithm for mapping an input value to an output value (hash value) with a certain length, the hash value has the characteristics of high randomness and collision resistance, the hash value may correspond to different input values, but it is impossible to find different input values within a normal operation time, and it is impossible to find two inputs with the identical hash value through a large number of operations.

Preferably, the hash can be regarded as an anti-counterfeit label of the input value, and the hash function is the anti-counterfeit labels of different videos.

Preferably, the camera module includes a camera, and the camera collects and segments a video to obtain video data.

Preferably, the block chain module includes a block chain, and the video data is subjected to camera hash calculation and packed transaction uplink through the block chain, and then is subjected to video hash on the chain.

Preferably, the behavior recognition module includes a behavior recognition service, the behavior recognition service requests for evidence storing information and obtains video data, the behavior recognition service performs cross-camera behavior analysis, the behavior recognition service links the used video hash and recognition result with evidence storing information, and then outputs result data.

The invention has the beneficial effects that:

1. the algorithm for identifying the behavior of people and carrying out early warning by cross-camera tracking through machine learning is adopted, the uniqueness and reliability of the video are determined through the Hash operation of the camera, the video is determined to be strictly increased in time sequence through a block chain, and no error is generated;

2. the blocks are connected through Hash, data in the blocks of the single chain are increased progressively according to the block height sequence, the video for indexing and storing the certificate is increased progressively according to the block shooting sequence, meanwhile, the occurrence time of the event can be determined according to the block timestamp, and the sequence of the event is guaranteed not to be disturbed.

3. The invention can improve the reliability of the result of the cross-camera behavior recognition and reduce the possibility that the behavior recognition result is questioned.

Drawings

FIG. 1 is a schematic diagram of the system of the present invention;

FIG. 2 is a schematic diagram of a first embodiment of the present invention;

FIG. 3 is a schematic view of a second embodiment of the present invention;

Detailed Description

In order to make the technical means, the original characteristics, the achieved purposes and the effects of the invention easily understood, the invention is further described below with reference to the specific embodiments and the attached drawings, but the following embodiments are only the preferred embodiments of the invention, and not all embodiments. Based on the embodiments in the implementation, other embodiments obtained by those skilled in the art without any creative efforts belong to the protection scope of the present invention.

Specific embodiments of the present invention are described below with reference to the accompanying drawings.

In the prior art, the event occurrence sequence is difficult to be determined, the camera time is based on the hardware local time, errors are easy to generate and easy to modify, and the accuracy of the camera time is difficult to guarantee. For some scenarios, such as incomplete camera coverage, or cyclic spatial continuous occurrence of events, this sequence is more easily confused.

Example 1

The first case is shown in fig. 2, assuming that the correct order of execution of events in 4 rooms is 1>2>3>4, if the middle walkway lacks camera coverage, the front-to-back order of the video is indeterminate, and if the walkway has camera coverage but any two rooms are not fully covered, the front-to-back order of the two rooms can be reversed.

Example 2

In another case, as shown in fig. 3, if the correct event execution sequence of 4 rooms is 1>2>3>4, but the operations are executed in the sequence of 2>3>4>1, the order of 1>2>3>4 can still be formed by changing the video sequence, because the algorithm itself cannot recognize the sequence, and the algorithm alone is easily questioned.

Example 3

As shown in fig. 1, a method for improving reliability of cross-camera behavior recognition by using a blockchain includes a camera module, a blockchain module, and a behavior recognition module, and includes the following steps:

the method comprises the following steps: defining a single-chain blockchain service;

step two: defining a camera Internet of things system which can automatically divide a video, calculate Hash to be stored as a file name, and send a signature to a block chain certificate after file Hash is packaged and traded;

step three: the camera service periodically polls the tile height and partitions the video according to the out-of-tile time of each tile in the chain of tiles.

The working principle is as follows: defining a single-chain block chain service, defining a camera Internet of things system capable of automatically dividing videos, calculating Hash to be stored as file names, and sending signatures of the Hash packed transaction of the files to a block chain evidence storage, wherein the camera service periodically polls the block height, divides the videos according to the block outlet time of each block of the block chain, the blocks are connected through the Hash, data in the blocks of the single chain are increased progressively according to the block height sequence, the video for indexing evidence storage is increased progressively according to the block outlet sequence, meanwhile, the time of event occurrence can be determined according to the block timestamp, and the sequence of events is ensured not to be disturbed.

Example 4

As shown in fig. 1, a method for improving reliability of cross-camera behavior recognition by using a blockchain includes a camera module, a blockchain module, and a behavior recognition module, and includes the following steps:

the method comprises the following steps: defining a single-chain blockchain service;

step two: defining a camera Internet of things system which can automatically divide a video, calculate Hash to be stored as a file name, and send a signature to a block chain certificate after file Hash is packaged and traded;

step three: the camera service periodically polls the tile height and partitions the video according to the out-of-tile time of each tile in the chain of tiles.

The blocks are connected through Hash, data in the blocks of the single chain are increased progressively according to the sequence of the block heights, the video with the index certificate is increased progressively according to the block shooting sequence, meanwhile, the occurrence time of events can be determined according to the block timestamps, the sequence of the events can be guaranteed not to be disturbed, the uniqueness and the reliability of the video are determined through Hash operation of the camera, and the video is determined to be strictly increased progressively according to the time sequence through the block chain.

The hash algorithm is an algorithm for mapping an input value to an output value (hash value) with a determined length, the hash value has the characteristics of high randomness and collision resistance, the hash value may correspond to different input values, but different input values cannot be found within normal operation time, it is impossible to find two inputs with the same hash value through a large number of operations, the hash value can be regarded as an anti-counterfeit label of the input value, and the hash function is an anti-counterfeit label of different videos.

The camera module includes the camera, the video is gathered and is cut apart to the camera, obtain video data, the block chain module includes the block chain, video data carries out camera calculation hash and packing transaction cochain through the block chain, then carry out video hash deposit certificate on the chain, the action recognition module includes the action recognition service, action recognition service requests deposit certificate information, and obtain video data, action recognition service strides camera action analysis, action recognition service will use video hash and identification result cochain deposit certificate, then output result data.

The working principle and the using method of the invention are as follows:

the working principle is as follows: based on an algorithm for performing cross-camera tracking through machine learning to identify personnel behaviors and perform early warning, the uniqueness and reliability of the video are determined through hash operation of the camera, and the video is determined to be strictly increased in time sequence through a block chain. The similar scheme of the invention is to utilize the time of watermark on the camera shooting picture, but the time of watermark is apt to produce the error; an algorithm for mapping an input value to a length-determined output value (hash value). The hash value has the characteristics of high randomness and collision resistance, so although the hash value may correspond to different input values, it is impossible to find different input values within normal operation time, and thus it can be considered impossible to find two inputs having the identical hash value through a large number of operations. On the basis, Hash can be taken as an anti-counterfeit label of an input value, the Hash in the text is taken as the anti-counterfeit label of different videos, blocks are connected through Hash, data in the blocks of a single chain are increased progressively according to the sequence of block heights, videos for indexing and storing certificates are increased progressively according to the sequence of the blocks, meanwhile, the time of occurrence of an event can be determined according to the block timestamp, and the sequence of the event is guaranteed not to be disordered.

The specific method comprises the following steps: the video is collected and divided by the camera to obtain video data, the video data is subjected to camera hash calculation and packed transaction chaining through a block chain, then video hash-on-chain storage is carried out, the behavior recognition module comprises behavior recognition service, the behavior recognition service requests storage certificate information and obtains the video data, the behavior recognition service carries out cross-camera behavior analysis, the behavior recognition service uses video hash and recognition result chaining storage certificates, and then result data are output.

The foregoing shows and describes the general principles, essential features, and advantages of the invention. It will be understood by those skilled in the art that the present invention is not limited to the embodiments described above, and the preferred embodiments of the present invention are described in the above embodiments and the description, and are not intended to limit the present invention. The scope of the invention is defined by the appended claims and equivalents thereof.