KR101677111B1 - Dynamic image object privacy protection device and the method of detecting the face of the pedestrian based - Google Patents

Abstract

A dynamic object image privacy protection apparatus based on the detection of a pedestrian's face and a method thereof are disclosed. The object of the present invention is to provide a dynamic object image privacy protection apparatus based on a pedestrian detection of a pedestrian, a server for storing an image, a frame extracting unit for selecting an image to be masked by the server and extracting a frame from the image, An object extraction unit for extracting an object by using an object detection algorithm for a frame in which the object is extracted and an authentication key using an image information of the object extracted by the object extraction unit and watermark data to be inserted, And a database (DB) for storing the authentication key generated in the masking unit, thereby protecting personal privacy.

Description

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a dynamic object image privacy protection apparatus and a method for protecting a dynamic object image privacy based on the detection of a pedestrian's face,

The present invention relates to a privacy protection apparatus, and more particularly, to a privacy protection apparatus for detecting a pedestrian's face based on pedestrian detection and face detection in an image, The present invention relates to a dynamic object image privacy protection apparatus and a method thereof.

CCTV is a system that transmits video signals from a video camera to a limited monitor and is used in places where surveillance is required, such as banks, airports, military facilities, and convenience stores. The CCTV may be used together with a DVR (Digital Video Recorder) to store images for a certain period of time and to store CCTV images only when there is a change in the video signals using various motion detection technologies.

The CCTV video signal is transmitted to a display device, an image processing device, or a storage device, and a still image frame is continuously used to display, process, or store the image. Here, the number of frames used per second for displaying, processing, or storing an image is expressed as FPS (Frames Per Second), and the minimum unit constituting an image of each frame is referred to as a pixel, The resolution is displayed by dividing the number of pixels by horizontal and vertical.

The FPS and the resolution are set according to the purpose of installation of the CCTV. As the FPS or the resolution of the processed image increases, the image processing speed of the computer or the image processing apparatus for processing the image becomes faster and the processing performance Of course.

We use a pedestrian detection algorithm and face recognition to extract objects from a human.

However, in the case of the pedestrian detection algorithm, the object can not be extracted when a part of the pedestrian comes out from the image or when it is imaged close.

In the case of the face recognition algorithm, the object can be extracted if it is imaged close, but when the distance is generated, the object can not be extracted properly.

KR Patent Publication No. 10-2014-0079502 (Qualcomm, June 26, 2014)

In order to solve such a problem, the present invention provides a dynamic object image privacy protection based on the detection of a pedestrian's face, which can protect personal privacy in an image by extracting an object to a person and masking the same using a simultaneous detection of a pedestrian and a face in an image And an object thereof is to provide an apparatus and a method therefor.

According to an aspect of the present invention, there is provided an apparatus for protecting a dynamic object image based on a pedestrian's face detection, the system comprising: a server for storing an image; a frame extracting unit for selecting an image to be masked by the server, An object extracting unit for extracting an object from the frame input by the frame extracting unit using an object detecting algorithm, and an image extracting unit for extracting an authentication information from the authentication information using the image information of the object extracted by the object extracting unit, And a database (DB) for storing an authentication key generated in the masking unit.

In addition, the masking unit can retrieve the watermarking by retrieving the object image input from the object extracting unit and the authentication key stored in the DB, and restore the image.

Meanwhile, a method for protecting a dynamic object image privacy based on the detection of a pedestrian's face includes the steps of (a) extracting a frame of an image to be encrypted in a server from a frame extracting unit, and (b) (C) extracting authentication information from an authentication information unit using image information of an object extracted by the object extraction unit and watermarking data to be inserted; and (D) storing the authentication key generated in the step (c) in a database (DB); and (e) generating a frame of the image to be restored from the image of the server (F) extracting an object from an object extraction unit by receiving an image of a frame extracted in the step (e), and (g) extracting an object extracted in the step (f) Person stored in DB To bring up and extracting the watermark from the masking part the key may include the step of reconstructing an image.

In step (b), the object extraction algorithm uses a face detection algorithm when it is determined that the object is close to the object, and extracts the object using the pedestrian detection algorithm when it is determined that the object is relatively far away.

Therefore, according to the apparatus and method for protecting a dynamic object image of a pedestrian based on the detection of a pedestrian of the present invention, personal privacy can be protected by simultaneously extracting a pedestrian object and masking it using both a pedestrian detection and a face detection in an image .

FIG. 1 is a block diagram of a dynamic object image privacy protection apparatus based on a face detection of a pedestrian according to an exemplary embodiment of the present invention. FIG.
FIG. 2 is a flowchart illustrating a method for protecting a dynamic object image privacy protection apparatus based on a face detection of a pedestrian according to an exemplary embodiment of the present invention;
FIG. 3 is a flowchart illustrating a masking method of a dynamic object image privacy protection apparatus based on a face detection of a pedestrian according to an embodiment of the present invention; FIG.
And,
4 is a block diagram of a device for restoring a dynamic object image based on a face detection of a pedestrian according to an embodiment of the present invention.

It is to be understood that the words or words used in the present specification and claims are not to be construed in a conventional or dictionary sense and that the inventor can properly define the concept of a term in order to describe its invention in the best possible way And should be construed in light of the meanings and concepts consistent with the technical idea of the present invention.

Throughout the specification, when an element is referred to as "comprising ", it means that it can include other elements as well, without excluding other elements unless specifically stated otherwise. It should be noted that the terms such as " part, "" module, " .

It is to be understood that the term "and / or" throughout the specification includes all possible combinations from one or more related items. For example, the meaning of "first item, second item and / or third item" may be presented from two or more of the first, second or third items as well as the first, second or third item It means a combination of all the items that can be.

Identification codes (e.g., a, b, c, ...) in each step throughout the specification are used for convenience of description, and the identification codes do not limit the order of each step, Unless the context clearly states a particular order, it may take place differently from the stated order. That is, each step may occur in the same order as described, may be performed substantially concurrently, or may be performed in reverse order.

Hereinafter, an apparatus for protecting a dynamic object image privacy based on the detection of a pedestrian's face according to an embodiment of the present invention will be described with reference to the drawings.

FIG. 1 is a block diagram of a dynamic object image privacy protection apparatus based on a face detection of a pedestrian according to an embodiment of the present invention. As shown in FIG. 1, The extracted object is extracted one frame at a time, the extracted object is masked, and the masked image is stored in the server 140. The encryption key used at this time is stored in the DB 150 and is restored to the original image The DB 150 calls an encryption key to release personal privacy and output an image.

To this end, the image privacy protection apparatus of the present invention includes a server 140 for storing an unencrypted image and an encrypted image, a frame 140 for selecting an image to be masked by the server 140, An object extraction unit 130 for extracting a pedestrian using an object extraction algorithm, a masking unit 110 for protecting an individual privacy by encrypting an object, and a database (DB) for storing an encryption key, (150).

The server 140 is configured as a security server for storing images, and stores a non-encrypted image and an encrypted private privacy image as a file.

Preferably, an unencrypted image, that is, a region for storing an original image, and an area for storing the encrypted image are distinguished from each other. When the encryption is performed, a list of unencrypted images is displayed. When decrypting, To be displayed.

The DB 150 stores the generated encryption key, and allows the corresponding encryption key to be used in decoding.

The frame extracting unit 160 selects an image to be encrypted and an image to be decrypted from the image stored in the server 140, and extracts the extracted image on a frame-by-frame basis.

That is, the server 140 is operated to fetch a file to be reproduced one frame at a time through the streaming service.

'Streaming service' refers to real-time playback of video, audio, and animation files on the Internet, which can shorten the playback time and limit the capacity of the drive because data can be implemented even before the file is transferred.

Accordingly, the frame extracting unit 160 can extract a frame from an image in real time even if the frame extracting unit 160 is connected to the server 140 via a network such as the Internet.

The object extracting unit 130 operates to extract an object or a pedestrian from a frame input from the frame extracting unit 160.

If it is determined that the object (pedestrian) is far away from the extracted frame, the dynamic object is extracted by the pedestrian detection algorithm. If it is determined that the object is nearby, the dynamic object is extracted using the face detection algorithm.

That is, in the present invention, when extracting a dynamic object according to the distance of an object, the object is extracted using a pedestrian detection algorithm and a face detection algorithm.

First, a pedestrian detection algorithm is a method of recognizing and tracking a pedestrian in real time using a Kalman filter and a clustering algorithm based on a hog continuing technique disclosed in KR Patent Publication No. 10-2015-0002040 d (2015.01.07) dp, Can be detected using the pedestrian detection and tracking method for intelligent video surveillance of patent No. 10-1588648 (2016.01.20).

Further, the face detection algorithm is disclosed in Korean Patent Laid-Open No. 10-2013-0112273 (Oct. 13, 2013), a description of a method and apparatus for acquiring a face image for face detection and recognition, or KR publication No. 10-2011-0103223 .20) using the SIFT algorithm for face recognition.

That is, in the present invention, since the face detector method and the pedestrian detection method are used in parallel, it is only necessary to extract the pedestrian detection and the face detection using the techniques disclosed in the paper or the like.

The object extracted by the object extraction unit 130 is encrypted in the masking unit 110 to protect personal privacy.

The masking unit 110 performs a masking process on the extracted dynamic object to protect personal privacy. The key is used to mask the dynamic object using the masking technique based on the chaotic system using the logistic map. The key used at this time is stored in the database, and when the original image is restored, the key stored in the database is fetched and the personal privacy is released.

For this, the masking unit 110 generates an authentication key using a Chaos-based zero-knowledge technique.

)에서 객체추출부(130)에서 객체추출알고리즘을 이용한 객체

를 추출하고, 수학식 1을 이용하여 인증키

를 생성한다. That is, the frame extracted by the frame extracting unit 160

The object extraction unit 130 extracts an object using an object extraction algorithm

And extracts the authentication key < RTI ID = 0.0 >

.

는 삽입할 워터마킹을,

는 카오스 기반의 Logistic map을 이용하여 생성되는 시퀀스이다.

A watermark to be inserted,

Is a sequence generated using a chaotic-based logistic map.

를 생성하기 때문의 고속의 연산이 가능하여 영상의 재생과 동시에 실시간으로 워터마킹 정보의 삽입과 추출이 이루어지는 것이다.The present invention performs an XOR operation,

So that the watermarking information can be inserted and extracted in real time simultaneously with the reproduction of the image.

Logistic map is one of chaos signal generators. It is a method of modeling growth rate R of population density.

If the value of R is less than 3.57, the output value X _{n + 1} of the current state has a linear characteristic. If the value of R is equal to or greater than 3.57, the output value of the present state is nonlinear by the quadratic X _n do.

The current state change value of the logistic map is used as a key value to encrypt with the frame data.

X ₁ (n), X ₂ (n), and X ₃ (n) are used as the authentication key of the system in applying a constant value representing the initial value and a pipeline type chaos algorithm for filling the blank of the clock in Equation 2.

This key represents the total length of 64 bits, each 16 bits. The number of key values generated according to the length of the image can be changed, and encryption is performed on two pixels at a time using the secret key. The encrypted data is transmitted in the form of encrypted image compressed data. The decryption process proceeds in the reverse order of the encryption process,

In addition, the masking unit 110 transmits the authentication key generated in the encryption process to the DB 150, stores the encrypted authentication key, and stores the encrypted image in the server 140.

The masking method of the present invention will be described using the above-described configuration.

)를 생성하고, "i=1"을 대입하여 인증키

을 생성한다(S132)Referring to FIG. 3, a method for masking a dynamic object image privacy protection based on a face detection of a pedestrian according to an embodiment of the present invention will be described. When a dynamic object is extracted in the object extraction unit 130 (S124) First, the masking unit 110 transmits an authentication key ("

), "I = 1" is substituted, and an authentication key

of (S132)

이 생성되면, 프레임추출부(160)에서 추출된 프레임(

)을 불러와 객체추출알고리즘을 이용하여 객체(

)을 추출한다(S133).In step S132,

The frame extracted by the frame extracting unit 160

) And extract the object (

(Step S133).

으로 동적 객체는

이 된다.Since i = 1 in step S112, the frame at this time is

As a dynamic object

.

을 생성하고, 카오스 기반의 Logistic map을 이용하여 시퀀스

을 생성한다. In addition, the masking portion 110 may include a water marking

And generates a sequence using a chaotic-based logistic map

.

을 생성한다(S134).Then, an XOR operation is performed using Equation (1)

(S134).

이 생성될 때까지 단계 S134와 S135을 반복하여 인증키

이 생성되면(S135), 마스킹부(110)는 생성된 인증키는 DB(150)에 저장하고, "i=i+1"을 대입하여,

를 생성하는 단계를 수행한다(S136).In step S135

Steps S134 and S135 are repeated until the authentication key

(S135), the masking unit 110 stores the generated authentication key in the DB 150, substitutes "i = i + 1"

(Step S136).

The generation of the authentication key is repeatedly generated for the set number of frames, and the generated authentication key is stored in the DB 150. [

A frame is a piece of picture that a video, a movie or a TV broadcasts on the screen when it delivers a video medium, and one piece of the picture is changed at a speed of about one second per second to produce a moving picture.

This makes it feel that the human vision is moving even after seeing one picture or one picture by showing another picture in the process of seeing one picture and transmitting it to the brain. It is a rate of how many frames (one picture) to show. It uses FPS (Frame Per Second) as a unit.

Usually, we have 24fps in the movies we see in the theater, and most of the videos used in the PCs are between 15fps and 30fps, and we have 29.97fps in Korea.

Therefore, since the number of frames (set value) varies depending on whether the original image is a video for what purpose, in step S137, it is determined whether i = set value?

When all the authentication keys set in step S137 are generated, the corresponding video is stored.

When the above-described steps S131 to S137 are repeated, the masked image is stored, the authentication key is stored in the server, and the stored authentication key is used for image restoration.

Hereinafter, a method for protecting the privacy protection apparatus will be described with reference to the drawings.

FIG. 2 is a flowchart illustrating a method for protecting a dynamic object image privacy protection apparatus based on a face detection of a pedestrian according to an embodiment of the present invention. As shown in FIG. 2, The security server 140 receives the image to be restored and extracts the frame (S110).

When the frame is extracted in step S110, the object extracting unit 130 extracts a dynamic object from the extracted frame (S120).

In step S120, the dynamic object is extracted, and the object is extracted using the face detection algorithm or the pedestrian detection algorithm according to the relative distance of the object.

More specifically, the object extraction unit 130 determines whether the object in the extracted frame is located in a short distance or a long distance.

The determination of such a distance can be made by a method capable of setting a remote area and a near area for each scenario at the time of image acquisition.

If it is determined in step S121 that the object is located at a short distance, a face detection algorithm is used (S122) to extract the object (S124). If it is determined in step S121 that there is an object at a relatively long distance, (S123), and the object is extracted (S124).

When the object is extracted in step S120, the masking unit 110 receiving the extracted object performs a masking process on the extracted object (S130)

This masking processing method can be referred to the process of FIG.

When the masking process is completed in step S130, the masking unit 110 stores the masked image in the server 140, and stores the encryption key (Ki) used for each frame in the DB 150 at this time.

In addition, the masked image stored in the server 140 and the encryption key stored in the DB 150 are used for image restoration.

Hereinafter, an apparatus for decrypting and reproducing an encrypted image will be described with reference to the drawings.

In the playback apparatus of the present invention, a file to be played back in the security server is fetched one frame at a time through a streaming service to extract the object, and the object is decoded and played back by each frame. The decryption key used at this time is called from the DB with the same key as the encryption key and is used for decryption.

Referring to a block diagram of a dynamic object image privacy restoration apparatus based on the detection of a pedestrian's face according to an embodiment of the present invention shown in FIG. 4, the server 140, in which the masked image is stored, And extracts the masked object from the object extraction unit 130. The masking unit 110 decodes the masked image using the encryption key stored in the DB 150 and outputs the decoded image .

The server 140, the frame extracting unit 160, and the DB 150 of the playback apparatus are the same as those in the storage apparatus, and thus a detailed description thereof will be omitted.

The masking unit 110 releases the personal privacy using the authentication key stored in the DB 150 using the Chaos-based zero-knowledge technique for the object input from the object extraction unit 130. [

Since the concrete decoding process is the same as the masking process described above, a detailed description thereof will be omitted.

As described above, the present invention can protect personal privacy in images by extracting and masking dynamic objects for a person using pedestrian detection and face detection.

While the present invention has been particularly shown and described with reference to exemplary embodiments thereof, it is evident that many alternatives, modifications and variations will be apparent to those skilled in the art.

110: masking unit 120: authentication information unit
130: object extracting unit 140:
150: DB 160: frame extracting unit

Claims (5)

delete A server for storing images;
A frame extracting unit for selecting an image to be masked by the server and extracting one frame through the streaming service from the corresponding image;
An object extraction unit for extracting an object using a face detection algorithm when it is determined that the object is located at a short distance from each frame input by the frame extraction unit and using a pedestrian detection algorithm when it is determined that the object is relatively far away;
A DB for storing the generated authentication key;
The authentication information generating unit generates an authentication key by calling the authentication information from the authentication information unit using the image information of the object extracted by the object extracting unit and the data of the watermark to be inserted and stores the generated authentication key in the DB, A masking unit for storing an extracted object and an authentication key stored in the DB to extract watermarking and restoring the image when the object is extracted from the image of the image to be restored in the server;
A device for protecting a dynamic object image privacy based on the detection of a pedestrian 's face. The method of claim 2,
The authentication key
An apparatus for protecting a dynamic object image privacy based on the detection of a pedestrian face represented by the following equation.

here

Lt; / RTI >

A watermark to be inserted,

The object extraction unit extracts the object using the object extraction algorithm,

Is a sequence generated using a chaotic-based logistic map.






delete delete

Images