TWI740145B - Method, electrical device, and computer program product for hiding information and protecting privacy of video - Google Patents

Abstract

A method for hiding information and protecting privacy of video is provided and includes: obtaining a video and performing an object detecting algorithm on a video frame to obtain at least one object area; replicating the object area to obtain recovery information, and embedding the recovery information into a non-object area of the video frame; embedding a authentication information into the object area and encrypting it to obtain an encrypted object area, and replacing the object area with the encrypted object area.

Description

Video information hiding and privacy protection methods, electronic Device, computer program product

The invention relates to a method for information hiding and privacy protection for specific objects in video.

With the development of technology and the reduction of hardware costs, video systems are widely used in all corners of life, such as mobile phones, driving recorders, network cameras and surveillance systems, etc., and therefore people may inadvertently violate The privacy of others. The video system is also often used as an important tool to record the whole incident. This also leads to people who want to tamper with the area of interest in the video. How to avoid the area of interest in the video from being tampered has become an important issue. Subject. In addition, information hiding is a technology that can protect and verify the authenticity of data. How to combine information hiding and privacy protection is a topic of concern to those skilled in the art.

An embodiment of the present invention provides a method for hiding and protecting privacy of video, which is suitable for an electronic device. The method includes: obtaining a video, executing an object detection algorithm on the video frame to obtain an object area; copying multiple copies of the object area to obtain Restore the information and embed the restored information in a non-object area of the video frame; and embed a verification information in the object area and encrypt it to obtain the encrypted object area, and replace the object area with the encrypted object area.

In some embodiments, the above method further includes: performing an object identification method on the object area, and if the object area does not belong to a privacy protection area, then the object area is excluded.

In some embodiments, the above method further includes: inputting the most significant bits of the pixels in the object area, the position information of the object area, and the number of the video frame into the hash function to obtain verification information.

In some embodiments, the aforementioned object area is copied at least 3 copies to obtain the restoration information, and the method further includes: determining whether the non-object area is sufficient to contain the restoration information; if the non-object area is insufficient to contain the restoration information, shrink the object area to obtain the restoration information Reduce the reduced object area, and copy at least 3 copies of the reduced object area to obtain restoration information.

In some embodiments, the aforementioned restoration information further includes the number of the video frame, compression information, the number of times the restoration information is hidden, the number of object regions, and the location information of the object regions.

In some embodiments, the aforementioned video frame includes a first video frame, a second video frame, and a third video frame, and the third video frame is between the first video frame and the second video frame. The method further includes: setting the filter mask on the timeline; identifying the object area in the video frame to obtain the matching of the object area between the video frames When applying a filter mask to the first video frame, if only the first video frame in the filter mask has an object area, delete the object area in the first video frame; and when the first video frame and the second video frame The distance between the video frames is less than or equal to a preset video frame distance, and both the first video frame and the second video frame have corresponding object areas, and the third video frame does not have an object area, filling the third video frame to make the third The video frame has a corresponding object area.

In some embodiments, the step of filling the third video frame includes: calculating the position of the object area in the third video frame according to the following equation (1).

X _i,j is the position of the j-th object area in the third video frame, L _j is the position of the j-th object area in the first video frame, R _j is the position of the j-th object area in the second video frame, i is the distance between the third video frame and the first video frame, and n is the number of video frames between the second video frame and the first video frame.

In some embodiments, the above method further includes: decrypting the encrypted object area to obtain the decrypted object area; obtaining information to be verified from the decrypted object area; judging whether the information to be verified is correct; if the information to be verified is correct, judging the encryption After that, the object area has not been tampered with; if the information to be verified is incorrect, restore the object area according to the restoration information.

From another perspective, the present invention provides an electronic device including a memory and a processor. The memory stores multiple instructions, and the processor executes these instructions to complete multiple steps: obtain video, execute an object detection algorithm on the video frame to obtain the object area; copy multiple copies of the object area to obtain the restoration information, and The restoration information is embedded in a non-object area of the video frame; A piece of authentication information is embedded in the object area and then encrypted to obtain the encrypted object area, and the object area is replaced with the encrypted object area.

From another perspective, the embodiment of the present invention provides a computer program product that is loaded and executed by an electronic device to complete multiple steps: obtain video, execute an object detection algorithm on the video frame to obtain the object area; Copy multiple copies of the object area to obtain recovery information, and embed the recovery information in a non-object area of the video frame; embed a verification information in the object area and encrypt it to obtain the encrypted object area, and replace the object area with the encrypted object area.

In order to make the above-mentioned features and advantages of the present invention more comprehensible, the following specific embodiments are described in detail in conjunction with the accompanying drawings.

100‧‧‧System

110‧‧‧Video capture module

120‧‧‧Electronic device

121‧‧‧Processor

122‧‧‧Memory

210‧‧‧Video

211~213、260‧‧‧Video frame

220, 242, 243, 252, 253‧‧‧ steps

230‧‧‧Database

221, 222‧‧‧Object area

241‧‧‧Verify Information

244‧‧‧Encrypted object area

251‧‧‧Non-object area

310, 320, 330‧‧‧Sequence

301‧‧‧Filter mask

311~316、321~323‧‧‧Video frame

410‧‧‧block

420‧‧‧Location information of the object area

430‧‧‧The number of the video frame

440‧‧‧Hash function

501~504‧‧‧Video frame

505‧‧‧Central area

601~609‧‧‧Step

610, 612‧‧‧Video frame

[Fig. 1] is a schematic diagram of a system according to an embodiment.

[FIG. 2A] and [FIG. 2B] are schematic diagrams illustrating the flow of information hiding and privacy protection methods according to an embodiment.

[Fig. 3] is a schematic diagram showing the operation of the filter mask according to an embodiment.

[Fig. 4] is a schematic diagram of calculating verification information according to an embodiment.

[Fig. 5] is a schematic diagram showing a video frame according to an embodiment.

[FIG. 6A] and [FIG. 6B] are flowcharts of verifying and replying to a video frame according to an embodiment.

Regarding the "first", "second", ... etc. used in this text, it does not specifically refer to the order or sequence, but only to distinguish the elements or operations described in the same technical terms.

Fig. 1 is a schematic diagram showing a system according to an embodiment. Please refer to FIG. 1, the system 100 here includes a video capture module 110 and an electronic device 120, where the electronic device 120 includes a processor 121 and a memory 122. The video capture module 110 may include a Charge-coupled Device (CCD) sensor, a Complementary Metal-Oxide Semiconductor (Complementary Metal-Oxide Semiconductor) sensor or other suitable photosensitive elements for capturing video . The electronic device 120 may be a smart phone, a tablet computer, a personal computer, a notebook computer, a server, an industrial computer, or various electronic devices with computing capabilities, etc. The present invention is not limited thereto. The processor 121 can be a central processing unit, a microprocessor, a microcontroller, a digital signal processor, an image processing chip or a special application integrated circuit, and the memory 122 can be a volatile memory or a non-volatile memory. In some embodiments, the video capture module 110 can also be provided in the electronic device 120, and the electronic device 120 can also be a monitor, a driving recorder, or the like. The memory 122 stores a plurality of instructions, and the processor 121 executes these instructions to complete a method for information hiding and privacy protection of the video. This method will be described in detail below.

2A and 2B are schematic diagrams illustrating the flow of information hiding and privacy protection methods according to an embodiment. 2A, first obtain a video 210. The video 210 includes a plurality of video frames 211 to 213. For each video frame, information hiding and privacy protection methods are performed. The following uses the video frame 211 as an example to continue the description.

In step 220, an object detection algorithm is executed to obtain object areas 221, 222 from the video frame 211. The object areas 221, 222 have corresponding bounding boxes for indicating the location of the object. In this embodiment, the aforementioned object detection algorithm is to detect human faces, but in other embodiments, it can also detect any objects such as license plates, certificates, pedestrians, and cars. In this embodiment, the object detection algorithm is a convolutional neural network, but in other embodiments, it can also be other machine learning algorithms such as support vector machines and adaboost.

In some embodiments, some human faces need privacy protection, while others do not. Therefore, these human faces that need privacy protection must be identified. Specifically, the database 230 stores the face images of multiple people. Here, an object recognition method can be implemented to determine whose face the object areas 221 and 222 belong to. The object area that needs to be protected is called Privacy protection area, if there is an object area that does not belong to the privacy protection area, then this object area is excluded. For example, the database 230 stores the faces of all residents in a community. The above object identification method is to determine whether the object areas 221 and 222 are residents. If they are, it is a privacy protection area, and subsequent privacy protection will continue. Method, if it is not a household, it will be excluded. For example, the object area 222 can be excluded, so the object area 221 will execute the subsequent steps. In some embodiments, the above-mentioned object recognition method may use any suitable algorithm such as a convolutional neural network, and the present invention is not limited thereto. In some embodiments, when the object detected in step 220 is a license plate, the above-mentioned object identification method is to identify the license plate number. The database 230 can also store the license plate number that requires privacy protection. If the object areas 221, 222 The license plate number in is not If it belongs to a specific license plate number, it means that it does not belong to a privacy protection area, and the subsequent steps will not be executed.

No matter which object detection method is used, there may be errors. These errors are divided into false positives and false negatives. In some embodiments, they can be eliminated according to the filter mask on the time axis. These errors. Specifically, please refer to FIG. 3, which is a schematic diagram illustrating the operation of the filter mask according to an embodiment. The sequence 310 represents the original video, and the video frames 311 to 316 represent the detected object area. The corresponding relationship between these object areas can be identified through the above object recognition method. For the sake of simplicity, the object areas in the video frames 311 to 316 belong to the same object (for example, the face of the same person). Here you can set a default video frame distance △ on the time axis, for example △=3, the size of the above filter mask is 2×△-1=5, including the current video frame, 2 front video frames and 2 When applying this filter mask, if only the current video frame has an object area in the filter mask, and the rest of the video frames have no object area, it is very likely that the object area in the current video frame is false positive. Therefore, this object area can be deleted. For example, when the filter mask 301 is applied to the video frame 312, since only the video frame 312 has an object area in the filter mask 301, this object area is deleted. Similarly, when the filter mask 301 is applied to the video frame 316, only the video frame 316 has an object area, so the object area in the video frame 316 is deleted. After applying the filter mask 301 to all video frames, the sequence 320 can be obtained.

Next, fill in the remaining video frames, if the distance between two video frames with object areas is less than or equal to the default video frame distance △, the video frame without the object area in the middle may be false negative, so all the video frames in the middle can be filled with the same object area. For example, the distance between the video frames 313 and 314 is 2, so the video frame 321 can be filled with the same object area. Similarly, the distance between the video frames 314 and 315 is 3, so the video frames 322 and 323 can be filled with the same object area, and the sequence 330 can be obtained after filling. From another perspective, if there are three video frames, they are called the first video frame, the second video frame, and the third video frame. The third video frame is between the first video frame and the second video frame. The distance between the first video frame and the second video frame is less than or equal to the preset video frame distance, and the distance between the second video frame and the third video frame is less than or equal to the preset video frame distance, and the first video frame and the second video frame are less than or equal to the preset video frame distance. Both video frames have an object area, and the third video frame does not have a corresponding object area, then the third video frame is filled so that the third video frame has a corresponding object area.

It is worth noting that the above step of filling the object area is not to modify the content of the video frame, but to add a bounding box to the video frame. Since the object area may move, the position of the newly added object area can be calculated by the method of inner difference. It is assumed here L, R is a video frame having two object region in a video frame needs to be filled between the video frame F _i L, R, I represents the distance between the video frame and video block F _i L, i = 1,2,...,n, n is the number of video frames between video frames L and R. The above filling can refer to the following equation (1).

Where X _i,j is the position of the j-th object area in the video frame F _i , L _j is the position of the j-th object area in the video frame L, R _j is the position of the j-th object area in the video frame R, Round () is the four arithmetic five-entry operations. For example, when the video frames 322 and 323 are to be filled, the video frame R in equation (1) is the video frame 314, the video frame L is the video frame 315, and n is 2, and when i=1, the video frame is filled. 322. When i=2, the video frame 323 is filled. The above-mentioned position may include the X and Y coordinates of the upper left and lower right coordinate points of the bounding box, a total of 4 values, and these 4 values can be applied to equation (1).

2B, in step 242, the verification information 241 is obtained, and the verification information 241 is embedded in the object area 221. FIG. 4 is a schematic diagram of calculating verification information according to an embodiment. 2B and FIG. 4, in this embodiment, the object area 221 is divided into a plurality of 2x4 pixel non-overlapping areas, such as area 410, and then capture the red, green, and blue of the 8 pixels in the area 410 The 7 most significant bits (MSB) of the grayscale value of the channel are 3x8x7=168 bits in total. In addition, the position information 420 of the object area 221 is obtained, such as the coordinates of the upper left and lower right points of the object area 221. In addition, the serial number 430 (also called serial number) of the video frame 211 is also obtained. The above-mentioned 168 bits, the position information 420 of the object area 221, and the number 430 of the video frame 211 are input into a hash function 440 to obtain the verification information 241. In some embodiments, the verification information 241 is 24 bits, and these bits are placed in a least significant bit of each gray scale value of the 8 pixels in the area 410, a total of 8×3×1=24 bits. For each 2x4 pixel area in the object area 221, the above steps are executed to complete the embedding of the verification information.

Returning to FIG. 2B, after the verification information 241 is embedded, in step 243, the object area 221 is encrypted to obtain the encrypted object area 244. This encryption algorithm can use Advanced Encryption Standard (AES) or other suitable encryption The secret algorithm, the present invention is not limited to this.

In addition, the non-object area 251 can be obtained from the video frame 211, where the object area 221 can be hidden to the non-object area 251. When the object area 221 is tampered with, information can be obtained from the non-object area 251 to restore the object area 221. First, the method of hiding image information is introduced. This method divides adjacent pixels in the image into multiple 1x2 non-overlapping blocks in pairs, and calculates the difference d of each group of pixels, which is expressed as the following equation (2).

d = g _i+1 - g _i …(2)

Where g _i represents the pixel at position i. The larger the absolute value of the difference d, it means that the block is in an edge area or a more complex image area, and vice versa, that the block is in a smooth area. The absolute value of the difference d corresponds to a quantization interval table to determine that this block can embed multi-bit confidential information. Please refer to Table 1 below. There are 6 intervals in Table 1. Each interval has a corresponding lower limit value, upper limit value and hidden value (number of bits). For example, the first interval has a lower limit value of 0 and an upper limit. Value 7, the hidden amount of this interval is 3 bits, and so on.

Using the upper limit value u _k and the lower limit value l _k of the k-th interval, k=1...6, the hidden amount C (unit is bit) can be calculated, as shown in the following equation (3).

After determining the amount of concealment, C bits can be obtained from the confidential information, denoted as b below. Through the lower limit value l _k and the confidential information b, the new difference value d ' of the block can be calculated, as shown in the following equation (4).

d ' = l _k + b …(4)

M then calculates the difference between the new difference value d 'and the original difference value d, as the following equation (5).

m=d ' -d…(5)

Finally, the confidential information is equally distributed to the two pixels of the block, and the adjusted pixel value can be expressed as the following equation (6).

,

)。舉例來說，如果g _i =35、g _i+1=45，則｜d｜=｜45-35｜=10，10屬於表1中的區間[8,15]，隱藏量為3位元，機密資訊b=5={101}₂。新的差值d'=8+5=13，而差值m=13-10=3，調整後的像素值分別為

、

。每個區塊在隱藏機密資訊前會檢查此區塊在嵌入最大可隱藏值時是否會溢位，若會溢位(例如，嵌入機密資訊以後的像素值大於255或小於0)則此區塊不做任何的嵌入動作。具體來說，可以測試f((g _i ,g _i+1),u_k-d)所計算出的兩個像素是否大於255或小於0，如果有其中一個像素大於255或小於0則判斷為溢位，這是因為u_k為對應區間的最大值，這會使計算出的像素(

,

)具有最大的差值。 Where f() represents a function to convert pixels ( g _i , g _{i +1} ) into pixels (

,

). For example, if g _i = 35 and g _{i +1} = 45, then |d|=|45-35|=10, 10 belongs to the interval [8,15] in Table 1, and the hidden amount is 3 bits. Confidential information b=5={101} ₂ . The new difference d ' =8+5=13, and the difference m=13-10=3, the adjusted pixel values are

,

. Before hiding confidential information, each block will check whether this block will overflow when embedding the maximum concealable value. If it will overflow (for example, the pixel value after embedding confidential information is greater than 255 or less than 0), then this block Do not do any embedding actions. Specifically, you can test whether the two pixels calculated by f (( g _i , g _{i +1} ), u _k - d ) are greater than 255 or less than 0. If one of the pixels is greater than 255 or less than 0, it is judged as Overflow, this is because u _k is the maximum value of the corresponding interval, which will make the calculated pixel (

,

) Has the largest difference.

In addition, the above equation (6) has the characteristics of the following equation (7).

f (( g _i , g _{i +1} ),m) = f ( f (( g _i , g _{i +1} ), m' ), m ") for m=m ' +m"...(7)

,

)，求出差值d*，如以下方程式(8)。 When you want to restore confidential information, first get the pixel (

,

) To find the difference d*, as shown in the following equation (8).

,

),u_k-d*)的結果，此結果會等同於f((g _i ,g _i+1),u_k-d)的結果，證明如以下方程式(9)。 Correspond the absolute value of the difference d _{* to the interval [l k} , u _k ] corresponding to the block found in Table 1. Next, check f ((

,

), u _k - d *), this result will be equivalent to the result of f (( g _i , g _{i +1} ), u _k - d ), which is proved as the following equation (9).

,

),u_k-d*)的結果顯示溢位，則表示原本的像素(g _i ,g _i+1)並沒有用來嵌入機密資訊，不用進行還原。如果f((

,

),u_k-d*)的結果並沒有溢位，根據以下方程式(10)可還原出機密資訊b。 Therefore, if f ((

,

), u _k - d *) shows overflow, which means that the original pixels ( g _i , g _{i +1} ) are not used to embed confidential information and do not need to be restored. If f ((

,

), the result of u _k - d *) does not overflow. According to the following equation (10), the confidential information b can be recovered.

、

的例子，首先計算出差值d*=47-34=13，查詢表1以後得到區間[8,15]，判斷f((34,47),15-13)並沒有溢位，因此最後可還原出機密資訊b=13-8=5={101}₂。 Continue the above

,

For example, first calculate the difference d*=47-34=13. After looking up the table 1, the interval [8,15] is obtained. It is judged that f ((34,47),15-13) does not overflow, so it can be finally Restore confidential information b=13-8=5={101} ₂ .

It is worth noting that the intervals and concealment amounts in Table 1 above can be modified appropriately. For example, Table 2 below presents different intervals and concealment amounts.

The concealment amount in Table 1 above is greater than the concealment amount in Table 2, but a large amount of concealment also means more image distortion. In some embodiments, different amounts of concealment can be set for different channels of pixels. For example, since the human eye is more sensitive to green, Table 2 can be used for the green channel, and Table 1 can be used for the blue and red channels. In some embodiments, red, green, and blue can be converted to different color spaces, such as YCbCr, and Table 2 with a smaller hidden amount is used for the luminance value Y, and the hidden amount is used for the chroma values Cb and Cr. Larger table 1. In other embodiments, other color spaces may also be used, and the present invention is not limited thereto.

The above-mentioned image information hiding method is only an example. In other embodiments, other image information hiding methods may also be used, and the present invention is not limited thereto. When other image information hiding methods are used, different hiding amounts can also be set in different channels.

Returning to Figure 2B, multiple copies of the object area 221 will be made here, and header information will be obtained in step 252 (the content will be described in detail below). These object areas 221 and the header information will form the restoration information. In step 253 The restoration information will be embedded in the non-object area 251. Specifically, in some embodiments, the header information is embedded in the two rows of pixels (also known as the video frame) on the upper, lower, left, and right sides of the video frame 211, as shown in FIG. 5 Shows that there are 4 video frames 501~504 in this video frame, and at least one piece of header information will be embedded in each video frame, which means that the header information will be embedded repeatedly at least 4 times to reduce the risk of header information being destroyed. . In addition, the object area 211 can be embedded in the central area 505 of the video frame except the video borders 501 to 504, and the object areas 221, 222. The object area 211 can also be embedded multiple times (at least 3 times, for example, 4 times). Reduce the risk of damage to the object area 211. In some embodiments, it is also determined whether the central area 505 is sufficient to accommodate 4 object areas 221. If the central area 505 is not enough to accommodate the 4 object areas 221, the object area 221 can be reduced to obtain the reduced object area. The reduced object area can be embedded in the central area 505. In some embodiments, wavelet transformation can be used. The low-frequency part is obtained to complete the reduction step, but in other embodiments, any suitable low-pass filter can also be used. In other embodiments, the header information can also be embedded in the central area 505, and the object area 221 can also be embedded in the video borders 501~504. That is to say, both the video borders 501~504 and the central area 505 can be collectively referred to as non-object areas. , The restoration information (including the header information and the object area 221) can be embedded in the non-object area in any order, position, and number of repetitions. The present invention does not limit the size, shape, location of the non-object area and the restoration information embedding The number of times. In some embodiments, before embedding the restoration information into the non-object area, any part (or all) of the restoration information can also be encrypted, for example, using the AES algorithm, thereby further protecting the restoration information. In some embodiments, the aforementioned header information includes the length of the header information, the number of the video frame 211, compression information (that is, whether the object area 211 has been reduced), the number of times the restoration information is hidden (that is, how many times the restoration information has been repeatedly embedded) ), object area The number and position information of the object area (for example, the coordinates of the upper left and lower right points of the bounding box).

Finally, the object area 221 is replaced with the encrypted object area 244 to obtain a protected video frame 260. The content of the object area 221 will not be seen in the video frame 260. In addition, the object area 221 can be verified and responded to. Specifically, FIG. 6A and FIG. 6B are flowcharts of verifying and replying to a video frame according to an embodiment. Please refer to Figure 2B, Figure 6A and Figure 6B. In step 601, the restoration information is obtained from the non-object area and the restoration information is restored. The steps of restoring a restoration information have been described in detail as the above equations (8)~(10). I will not repeat them here. It is assumed here that n pieces of restoration information have been obtained, expressed as R _i ,i=1,2,...,n. Next, a majority vote can be used to vote on each bit to generate the final recovery information, so that the b-th bit in the recovery information is ER ^b , then ER ^b can be expressed as the following equation (11).

為第i份還原資訊中的第b個位元。在一些實施例中，還原資訊在嵌入至非物件區域之前有經過加密，因此在步驟601也需要經過解密才可以得到還原資訊。如果物件區域要先前有經過縮小的處理，則在步驟601需要將縮小後物件區域放大，在此可以採用任意合適的內差方法(interpolation)，例如雙線性內插法(bilinear interpolation)。 in

Restore the b-th bit in the information for the i-th copy. In some embodiments, the restoration information is encrypted before being embedded in the non-object area. Therefore, decryption is also required in step 601 to obtain the restoration information. If the object area needs to be reduced previously, then in step 601, the reduced object area needs to be enlarged. Here, any suitable interpolation method, such as bilinear interpolation, can be used.

In step 602, it is determined whether the number of the video frame is correct, for example, it is checked whether the number of the current video frame is continuous with the previous video frame and the next video frame. If the result of step 602 is otherwise, proceed to the next step in step 603 If yes, go to step 604. In step 604, it is determined whether the encoding time of the video is correct, for example, it is checked whether the encoding time of the current video frame is continuous with the previous video frame and the next video frame. If the result of step 604 is otherwise, proceed to step 603, if yes, proceed to step 605. In step 605, determine whether there is an object area in the video frame 260 according to the header information, if not, proceed to step 603, and if yes, proceed to step 606.

In step 606, the encrypted object area 244 is decrypted according to the AES algorithm to obtain the decrypted object area. In step 607, obtain a piece of information to be verified from the decrypted object area, that is, repeat the steps in Figure 4, divide the decrypted object area into 2x4 pixel blocks, and obtain the 7 most effective pixels for each pixel in this block Bit, obtain the position information of the decrypted object area and the number of the video frame from the header information, and finally calculate a piece of information to be verified through the hash function. In step 608, it is determined whether the information to be verified is correct, specifically, the verification information 241 is obtained from the least significant bit of the pixel in the decrypted object area, and it is determined whether the verification information 241 is the same as the information to be verified calculated in step 607 If the same means that the verification information is correct (the encrypted object area 244 has not been tampered with), then proceed to step 609 to replace the encrypted object area 244 with the decrypted object area, and finally a video frame 610 is obtained. If the result of step 608 is no, it means that the encrypted object area 244 has been tampered with. In step 611, the restored object area is pasted back to the video frame 260 according to the position of the object area in the header information, and finally the video frame 612 can be obtained. .

From another perspective, the present invention also proposes a computer program product, which can be written in any programming language and/or platform. When the computer program product is loaded into the computer system and executed, it can execute the above The method of Fig. 2A, Fig. 2B, Fig. 6A, and Fig. 6B.

Although the present invention has been disclosed in the above embodiments, it is not intended to limit the present invention. Anyone with ordinary knowledge in the relevant technical field can make some changes and modifications without departing from the spirit and scope of the present invention. The protection scope of the present invention shall be subject to those defined by the attached patent application scope.

260‧‧‧Video Frame

242, 243, 252, 253‧‧‧ steps

221‧‧‧Object area

241‧‧‧Verify Information

244‧‧‧Encrypted object area

251‧‧‧Non-object area

Claims (9)

A video information hiding and privacy protection method is applicable to an electronic device. The method includes: obtaining a video, executing an object detection algorithm on at least one video frame of the video to obtain at least one object area; Copy at least 3 copies of the object area to obtain multiple pieces of restored information, determine whether a non-object area of the at least one video frame is sufficient to accommodate the restored information, if otherwise, reduce the at least one object area to obtain at least one reduced object area, And copy at least 3 copies of the at least one reduced object area to obtain the restoration information; embed the restoration information in the non-object area; and embed verification information in the at least one object area and encrypt it to obtain at least one encryption And replace the at least one object area with the at least one encrypted object area. For example, the method described in claim 1 further includes: performing an object identification method on at least one object area, and if the at least one object area does not belong to a privacy protection area, excluding the at least one object area. For example, the method described in item 1 of the scope of the patent application further includes: the most significant bits of the pixels in the at least one object area Yuan, the location information of the at least one object area, and a number of the at least one video frame are input to a hash function to obtain the verification information. For the method described in item 1 of the scope of patent application, each of the restored information further includes the number of the at least one video frame, a compressed information, a number of times that the restored information is hidden, the number of the at least one object area, and the at least one The location information of the object area. For the method described in item 1 of the scope of patent application, wherein the number of the at least one video frame is greater than 1, the video frames include a first video frame, a second video frame, and a third video frame, and the third video frame The frame is between the first video frame and the second video frame, and the method further includes: setting a filter mask on a time axis; identifying the at least one object area in each of the video frames to obtain the The corresponding relationship between the at least one object area among the video frames; when the filter mask is applied to the first video frame, if only the first video frame in the filter mask has the at least one object area, delete The at least one object area in the first video frame; and when the distance between the first video frame and the second video frame is less than or equal to a preset video frame distance, and the first video frame and the second video frame The frames all have the corresponding at least one object area, and the third video frame does not have the at least one object area, filling the third video frame so that the third video frame has the corresponding at least one object area. According to the method described in claim 5, the step of filling the third video frame includes: calculating the position of the at least one object area in the third video frame according to the following equation (1):

Where X _i,j is the position of the j-th object area in the at least one object area in the third video frame, L _j is the position of the j-th object area in the at least one object area in the first video frame, R _{j is} the position of the j-th object area in the at least one object area in the second video frame, i is the distance between the third video frame and the first video frame, and n is the second video frame and the The number of video frames between the first video frames. For example, the method described in item 1 of the scope of patent application further includes: decrypting the at least one encrypted object area to obtain at least one decrypted object area; obtaining information to be verified from the at least one decrypted object area; determining the to-be-verified Whether the information is correct; if the information to be verified is correct, it is determined that the at least one encrypted object area has not been tampered with; and if the information to be verified is incorrect, the at least one object area is restored based on the restored information. An electronic device, including: A memory storing a plurality of instructions; and a processor executing the instructions to complete a plurality of steps: obtaining a video, and executing an object detection algorithm on at least one video frame of the video to obtain at least one object area ; Copy at least 3 copies of the at least one object area to obtain multiple pieces of restoration information, determine whether a non-object area of the at least one video frame is sufficient to accommodate the restoration information, if not, reduce the at least one object area to obtain at least one Reduce the reduced object area, and copy at least 3 copies of the at least one reduced object area to obtain the restoration information; embed the restoration information in the non-object area; and embed verification information into the at least one object area and then encrypt To obtain at least one encrypted object area, and replace the at least one object area with the at least one encrypted object area. A computer program product loaded and executed by an electronic device to complete multiple steps: obtaining a video, executing an object detection algorithm on at least one video frame of the video to obtain at least one object area; and the at least one object Copy at least 3 copies of the area to obtain multiple pieces of restored information, determine whether a non-object area of the at least one video frame is sufficient to accommodate the restored information, if not, reduce the at least one object area to obtain at least one reduced object area, and Copy at least 3 copies of the at least one reduced object area to obtain the restoration information; embed the restoration information in the non-object area; and Embedding a piece of verification information into the at least one object area and encrypting it to obtain at least one encrypted object area, and replacing the at least one object area with the at least one encrypted object area.

Images