CN117710497A

CN117710497A - Image anonymizing method

Info

Publication number: CN117710497A
Application number: CN202211104327.4A
Authority: CN
Inventors: 杜伟伟; 陈丹; 鲁梦琦; 卢晨光
Original assignee: Valeo Interior Controls Shenzhen Co Ltd
Current assignee: Valeo Interior Controls Shenzhen Co Ltd
Priority date: 2022-09-09
Filing date: 2022-09-09
Publication date: 2024-03-15

Abstract

The present disclosure provides an image anonymizing method, an image anonymizing apparatus and device, a computer-readable storage medium, and a computer program product. The image anonymization method comprises the following steps: determining a region to be anonymized of the image; determining a plurality of layers to be anonymized according to the size of the region to be anonymized; determining an area and anonymization parameters of each of a plurality of layers; and anonymizing the image based on the area of each layer and anonymizing parameters for each layer of the plurality of layers.

Description

Image anonymizing method

Technical Field

The present disclosure relates to the field of computer technology, and more particularly, to an image anonymizing method, an image anonymizing apparatus and device, a computer-readable storage medium, and a computer program product.

Background

Computer Vision (CV) refers to the operation of identifying, tracking, measuring, etc. objects by using a camera and a Computer instead of human eyes, and the main task of the CV is to obtain three-dimensional information of a corresponding scene by processing an acquired picture or video. Computer vision is now widely used in many fields such as industrial control, medical imaging diagnosis, video surveillance, face recognition, driving assistance, and the like. For example, in the automotive field, by performing image acquisition and processing on pedestrians, road conditions, and the like using computer vision, development and application of functions such as three-dimensional visual navigation and automatic driving can be facilitated.

On the other hand, with the continuous development of information technology, information security has become more important. The governments of various countries are orderly put out of data security regulations, and require anonymization processing of privacy information such as faces, license plates and the like acquired in scene capturing, which brings new challenges to computer vision algorithms. First, for true anonymization purposes, image anonymization should be irreversible; secondly, the image anonymization processing should not introduce new texture features to the image so as not to influence the CV algorithm; finally, since the privacy information to be anonymized includes features of different categories such as faces and license plates, it is desirable that the image anonymizing method can be compatible with the features of the different categories. However, existing methods of image anonymization, such as generating dummy data to cover real data by antagonizing a network, monochromatically filling anonymized areas, single-layer gaussian blur, etc., cannot simultaneously satisfy the above requirements.

Disclosure of Invention

In order to effectively solve the above-described problems, the present disclosure provides an image anonymizing method, an image anonymizing apparatus and device, a computer-readable storage medium, and a computer program product.

According to an aspect of the embodiments of the present disclosure, there is provided an image anonymizing method, including: determining a region to be anonymized of the image; determining a plurality of layers to be anonymized according to the size of the region to be anonymized; determining an area and anonymization parameters of each of the plurality of layers; and anonymizing the image for each of the plurality of layers based on an area of the layer and anonymizing parameters.

According to an example of an embodiment of the present disclosure, determining a plurality of layers to be anonymized according to a size of the region to be anonymized includes: the larger the proportion of the region to be anonymized in the image is, the larger the number of layers to be anonymized is determined.

According to an example of an embodiment of the present disclosure, wherein determining the area and anonymization parameters of each of the plurality of layers comprises: sequentially increasing the area of each layer in the plurality of layers from the center to the periphery of the region to be anonymized; and sequentially decreasing anonymization parameters of each layer of the plurality of layers from the center to the periphery of the region to be anonymized.

According to an example of an embodiment of the present disclosure, wherein determining the area and anonymization parameters of each of the plurality of layers comprises: the side length or radius of each layer of the plurality of layers is made a predetermined number of pixels greater than the side length or radius of the previous layer.

According to an example of an embodiment of the present disclosure, the predetermined number of pixels is 4 to 10 pixels.

According to an example of an embodiment of the present disclosure, wherein determining the area and anonymization parameters of each of the plurality of layers comprises: the edge length difference or the radius difference of adjacent layers in the plurality of layers is made equal to a predetermined multiple of the difference of anonymization parameters of the adjacent layers.

According to an example of an embodiment of the present disclosure, the predetermined multiple is 2 times.

According to an example of an embodiment of the present disclosure, wherein determining the area and anonymization parameters of each of the plurality of layers comprises: determining the area of the first layer and anonymization parameters based on the size of the region to be anonymized; and determining the area and anonymization parameters of the rest layers in the layers according to the area and anonymization parameters of the first layer.

According to an example of an embodiment of the disclosure, wherein the area of the first layer is smaller than or equal to the area of the region to be anonymized, and wherein determining the area and anonymization parameters of the remaining layers of the plurality of layers according to the area and anonymization parameters of the first layer comprises: sequentially increasing the area of the rest of the plurality of layers compared to the area of the first layer; and sequentially decrementing anonymization parameters of remaining layers of the plurality of layers compared to the anonymization parameters of the first layer.

According to an example of an embodiment of the disclosure, the anonymization process is gaussian blur, and the anonymization parameter is a gaussian blur radius.

According to another aspect of the embodiments of the present disclosure, there is provided an image anonymizing apparatus, the apparatus including: a region determining unit configured to determine a region to be anonymized of the image; a layer determining unit configured to determine a plurality of layers to be anonymized according to the size of the region to be anonymized, and determine an area and anonymization parameters of each layer in the plurality of layers; and a processing unit configured to anonymize the image based on an area of the layer and anonymization parameters for each layer of the plurality of layers.

According to an example of an embodiment of the present disclosure, the area determination unit is further configured to: the larger the area to be anonymized is, the larger the number of layers to be anonymized is determined.

According to an example of an embodiment of the disclosure, the layer determining unit is further configured to: sequentially increasing the area of each layer in the plurality of layers from the center to the periphery of the region to be anonymized; and sequentially decreasing anonymization parameters of each layer of the plurality of layers from the center to the periphery of the region to be anonymized.

According to an example of an embodiment of the disclosure, the layer determining unit is further configured to: the side length or radius of each layer of the plurality of layers is made a predetermined number of pixels greater than the side length or radius of the previous layer.

According to an example of an embodiment of the disclosure, the layer determining unit is further configured to: the edge length difference or the radius difference of adjacent layers in the plurality of layers is made equal to a predetermined multiple of the difference of anonymization parameters of the adjacent layers.

According to an example of an embodiment of the disclosure, the layer determining unit is further configured to: determining the area of the first layer and anonymization parameters based on the size of the region to be anonymized; and determining the area and anonymization parameters of the rest layers in the layers according to the area and anonymization parameters of the first layer.

According to an example of an embodiment of the present disclosure, wherein the area of the first layer is smaller than or equal to the area of the region to be anonymized, and wherein the layer determining unit is further configured to: sequentially increasing the area of the rest of the plurality of layers compared to the area of the first layer; and sequentially decrementing anonymization parameters of remaining layers of the plurality of layers compared to the anonymization parameters of the first layer.

According to another aspect of the embodiments of the present disclosure, there is provided an image anonymizing apparatus including: one or more processors; and one or more memories, wherein the memories have stored therein computer readable instructions that, when executed by the one or more processors, cause the one or more processors to perform the method of any of the various aspects of the disclosure.

According to another aspect of embodiments of the present disclosure, there is provided a computer-readable storage medium having stored thereon computer-readable instructions, which when executed by a processor, cause the processor to perform a method according to any of the above aspects of the present disclosure.

According to another aspect of embodiments of the present disclosure, there is provided a computer program product comprising computer readable instructions which, when executed by a processor, cause the processor to perform a method as in any of the above aspects of the present disclosure.

With the image anonymizing method, the image anonymizing apparatus and the device, the computer readable storage medium and the computer program product according to the above aspects of the embodiments of the present disclosure, by performing image anonymization processing based on a plurality of layers, it is possible to ensure that there is a step transition region between different layers while performing irreversible anonymization on a region to be anonymized, thereby avoiding edge features at edges of different layers, and being capable of better compatible with anonymization features of different categories, for example, better anonymizing a face region and a license plate region in an image at the same time.

Drawings

The above and other objects, features and advantages of the presently disclosed embodiments will become more apparent from the more detailed description of the presently disclosed embodiments when taken in conjunction with the accompanying drawings. The accompanying drawings are included to provide a further understanding of embodiments of the disclosure, and are incorporated in and constitute a part of this specification, illustrate embodiments of the disclosure and together with the description serve to explain the disclosure, without limitation to the disclosure. In the drawings, like reference numerals generally refer to like parts or steps.

FIG. 1 illustrates a flow chart of an image anonymization method according to embodiments of the present disclosure;

FIG. 2 illustrates an example of multiple layers according to an example of an embodiment of the present disclosure;

FIG. 3 illustrates an effect contrast diagram of an example image anonymization method according to embodiments of the present disclosure;

FIG. 4 illustrates an effect contrast diagram of another example image anonymization method according to embodiments of the present disclosure;

fig. 5 shows a schematic configuration diagram of an image anonymizing apparatus according to an embodiment of the present disclosure;

fig. 6 illustrates a schematic diagram of an architecture of an exemplary computing device, according to an embodiment of the present disclosure.

Detailed Description

The technical solutions in the embodiments of the present disclosure will be clearly and completely described below with reference to the drawings in the embodiments of the present disclosure. It will be apparent that the described embodiments are merely embodiments of a portion, but not all, of the present disclosure. All other embodiments, which can be made by one of ordinary skill in the art without the need for inventive faculty, are intended to be within the scope of the present disclosure, based on the embodiments in this disclosure.

In computer vision applications, it is desirable that the image anonymization algorithm can meet the following requirements. First, in order to achieve the real anonymization purpose, the image anonymization should be irreversible, that is, the anonymized information such as the face, the license plate and the like cannot be restored by the image restoration and the like, which requires that the image anonymization process has enough intensity; secondly, no new texture features should be introduced into the image during the image anonymization process to avoid affecting the CV algorithm, and the anonymized image should be as natural as possible visually, which requires that the image anonymization process should not be too strong, otherwise significant edge features will be generated at the edges of the anonymized region; third, since privacy information to be anonymized includes features of different categories such as faces and license plates, it is desirable that the image anonymizing method can be compatible with the features of the different categories at the same time.

In the existing image anonymization algorithm, the generation countermeasure network algorithm achieves the face anonymization by generating the virtual face to cover the original face features, however, the algorithm requires stronger computing processing capacity and needs to operate based on RGB images, so that the original image format data cannot be directly processed. In addition, the method does not support anonymization processing of license plates, which greatly limits the application range of generating an countermeasure network algorithm.

The method is characterized in that the method comprises the steps of carrying out a color block filling operation on a color block to be anonymized, wherein the color block filling operation is carried out on the color block to be anonymized, and the color block filling operation is carried out on the color block to be anonymized.

Another more common image anonymization algorithm is single-layer gaussian blur. Gaussian blur, also known as gaussian smoothing, reduces image noise and reduces the level of detail by convolving the image with a normal distribution. The single-layer Gaussian blur is to directly perform one or more Gaussian blur processes on the image area to be anonymized, and the single-layer Gaussian blur can also anonymize both faces and license plates, but the smoothness of the edges of the image anonymizing area cannot be ensured, namely edge features are introduced, and particularly, the stronger edge features are generated under the condition of stronger Gaussian blur processes for realizing irreversible anonymization.

The image anonymizing method can achieve irreversible image anonymization without introducing new texture features, and can be well compatible with different types of features to be anonymized such as faces, license plates and the like.

An image anonymization method according to an embodiment of the present disclosure is described below with reference to fig. 1. Fig. 1 shows a flowchart of an image anonymization method 100 according to embodiments of the present disclosure. As shown in fig. 1, in step S110, a region of an image to be anonymized is determined. In the embodiment of the disclosure, an image refers to an image acquired by a camera or a picture frame in a video, and a region to be anonymized of the image refers to a region to be anonymized of the image. For example, if anonymization processing is desired for a face appearing in an image, the region to be anonymized refers to a face region in the image. For another example, if it is desired to anonymize a license plate of an automobile appearing in an image, the region to be anonymized refers to a license plate region in the image. It should be noted that, although the face and the license plate are taken as examples of the region to be anonymized above, the embodiment of the present disclosure is not limited thereto, and the region to be anonymized in the embodiment of the present disclosure may be any image region that needs to be anonymized.

In step S120, a plurality of layers to be anonymized are determined according to the size of the region to be anonymized. Specifically, in the embodiment of the present disclosure, multiple layers may be determined in an image according to the size of a region to be anonymized, and anonymization processing may be performed for each layer in the multiple layers, so as to achieve a better anonymization effect. In the embodiment of the present disclosure, the plurality of layers may be two layers, three layers, or more layers, and the number of layers is not particularly limited in the embodiment of the present disclosure.

Here, the anonymization process is described by taking gaussian blur as an example. In this example, the image anonymization method based on multiple layers according to the embodiments of the present disclosure may be referred to as multi-layer gaussian blur, as compared to the conventional single-layer gaussian blur method for only a single layer. In general, a single-layer gaussian blur performs anonymization processing only on a region to be anonymized (i.e., a single layer) in an image, and an image anonymization method according to an embodiment of the present disclosure, or more specifically, a multi-layer gaussian blur in this example, determines a plurality of layers based on the region to be anonymized, and then performs anonymization processing for each of the plurality of layers. Although the description is given here by taking gaussian blur as an example, the embodiments of the present disclosure are not limited thereto, and anonymization processing based on a plurality of layers in the present disclosure may be performed by using other suitable image anonymization techniques.

According to an example of an embodiment of the present disclosure, the greater the proportion of the region to be anonymized in the image, the greater the number of layers to be anonymized is determined. In general, the larger the proportion of the region to be anonymized in the image, the higher the resolution of the region to be anonymized or the more image details are contained, so that stronger anonymization processing is required to realize irreversible anonymization. For example, if the tail of an automobile is photographed at a short distance, so that the license plate area in the obtained image occupies a relatively large area, the resolution of the license plate number in the image will be very high, in which case, in order to achieve irreversible anonymization of the license plate, a stronger anonymization process needs to be performed on the license plate. At this time, if the anonymization is performed by using the conventional single-layer gaussian blur method, a large gaussian blur radius is required or the same region is consecutively gaussian blurred a plurality of times, which results in very significant edge features at the edge of the license plate. In contrast, according to the image anonymizing method of the embodiment of the present disclosure, by determining a plurality of layers based on the region to be anonymized and anonymizing the images layer by layer, irreversible anonymization can be achieved without introducing edge features, as will be described in further detail below.

After the plurality of layers to be anonymized are determined, in step S130, the area of each of the plurality of layers and the anonymization parameter are determined. In examples of embodiments of the present disclosure, each of the plurality of layers may be a square area, a circular area, a polygonal area, etc., to which embodiments of the present disclosure are not particularly limited. In an example of an embodiment of the present disclosure, the anonymization parameter is a parameter employed when anonymizing, and its size reflects the strength of the anonymization. For example, in the case where the image anonymization method according to the embodiment of the present disclosure is based on gaussian blur, the anonymization parameter refers to a gaussian blur radius or a so-called gaussian kernel, and this value represents, for any pixel to be processed, how many surrounding values of pixels are employed for smoothing processing, and the larger the blur radius, the more blurred the processed image.

According to an example of an embodiment of the present disclosure, the area of each of the plurality of layers may be sequentially increased from the center to the periphery of the region to be anonymized, and the anonymization parameter of each of the plurality of layers may be sequentially decreased from the center to the periphery of the region to be anonymized. That is, the layer closest to the center of the region to be anonymized has the smallest area and the anonymization parameter is the largest. Taking fig. 2 as an example for illustration, fig. 2 shows an example of multiple layers according to an example of an embodiment of the present disclosure. In fig. 2, the hatched portion indicates the region to be anonymized, and the square region from the center to the periphery of the region to be anonymized is layer 1, layer 2, layer 3, and layer 4 in this order. It can be seen that from layer 1 to layer 4, the area of the layers increases gradually, but the anonymization parameters applied to layer 1 to layer 4 may decrease gradually. Taking anonymization parameter as gaussian blur radius as an example, the gaussian blur radii of layer 1 to layer 4 may be respectively: 15. 11, 7, 5, namely 15 pixels, 11 pixels, 7 pixels and 5 pixels, respectively.

In an example of an embodiment of the present disclosure, the area of the first layer and anonymization parameters may be determined first based on the size of the region to be anonymized. For example, the area of the first layer may be made smaller than or equal to the area of the region to be anonymized. For example, as shown in fig. 2, in the case where the area to be anonymized has been determined, it may be determined that the area of the first layer is equal to the area of the area to be anonymized, that is, it is determined that the first layer is layer 1, and anonymization parameters of the first layer may be determined, for example, it is determined that the anonymization parameters of layer 1 are 15. And then, determining the area and anonymization parameters of the rest layers in the layers according to the area and anonymization parameters of the first layer. Specifically, as described above, the areas of the remaining layers may be sequentially increased compared to the area of the first layer, for example, the remaining layers may be determined to be layer 2, layer 3, and layer 4, which are sequentially increased in area; and the anonymization parameters of the rest layers are sequentially decreased compared with the anonymization parameters of the first layer, for example, the gaussian blur radii of the layers 2 to 4 may be respectively: 11. 7, 5.

It should be noted that, although 4 layers are described above in connection with fig. 2 as an example, the embodiments of the disclosure are not limited thereto, and as described above, the number of layers may be more or less according to the size of the region to be anonymized and the actual requirement. In addition, in fig. 2, the area to be anonymized and each layer are exemplified as square areas, but the embodiments of the present disclosure are not limited thereto, and the area to be anonymized may be any possible shape according to the actual anonymization requirement, and each layer may take any suitable shape, such as rectangle, circle, and the like.

Further, according to examples of embodiments of the present disclosure, each layer of the plurality of layers may be made to have a side length or radius that is a predetermined number of pixels greater than the side length of the previous layer. For example, in the case where each layer is a square area, the side length of each layer may be made larger than the side length of the previous layer by a predetermined number of pixels. For another example, in the case where each layer is a circular area, the radius of each layer may be made a predetermined number of pixels larger than the radius of the previous layer. The predetermined number of pixels may be, for example, 4 to 10 pixels, but the embodiment of the present disclosure is not limited thereto, and the specific value of the predetermined number may be determined according to actual needs. For example, in fig. 2, the side length of layer 2 may be made 10 pixels longer than the side length of layer 1, the side length of layer 3 may be made 10 pixels longer than the side length of layer 1, and the side length of layer 4 may be made 10 pixels longer than the side length of layer 1.

Alternatively, according to examples of embodiments of the present disclosure, the edge length difference or the radius difference of adjacent layers of the plurality of layers may be made equal to a predetermined multiple of the difference in anonymization parameters of the adjacent layers. For example, in the case where each layer is a square region, the difference in side length of adjacent layers among the plurality of layers may be made equal to a predetermined multiple of the difference in anonymization parameters of the adjacent layers. For another example, in the case where each layer is a circular area, the difference in radius of adjacent layers among the plurality of layers may be made equal to a predetermined multiple of the difference in anonymization parameters of the adjacent layers. In the example of the embodiment of the present disclosure, the predetermined multiple may be, for example, 2 times, but the embodiment of the present disclosure is not limited thereto, and a specific value of the predetermined multiple may be determined according to actual requirements.

In this example, the area of the first layer and anonymization parameters of each layer may be first determined, and then the areas of the remaining layers may be determined based on the area of the first layer and anonymization parameters of each layer. Still taking fig. 2 as an example, assuming that the side length of the square layer 1 is 50 pixels, the gaussian blur radii of the layers 1 to 4 are respectively: 15. 11, 7, 5, the side length of layer 2 can be determined to be 58 pixels by making the side length difference of layer 2 and layer 1 2 be 2 times the blur radius difference of layer 2 and layer 1, and the area of layer 2 is larger than layer 1; similarly, the side lengths of layer 3 and layer 4 may be determined to be 66 pixels and 70 pixels, respectively.

Thereafter, in step S140, for each of the plurality of layers, an anonymization process is performed on the image based on the area of the layer and the anonymization parameter. For example, in the example of fig. 2, assume that the gaussian blur radii of layers 1 to 4 are respectively: 15. 11, 7, 5, then the layer 1 may be sequentially subjected to a gaussian blur with a blur radius of 15 pixels, the layer 2 may be subjected to a gaussian blur with a blur radius of 11 pixels, the layer 3 may be subjected to a gaussian blur with a blur radius of 7 pixels, and the layer 4 may be subjected to a gaussian blur with a blur radius of 5 pixels.

To better illustrate the principles and advantages of the image anonymization method according to embodiments of the present disclosure, further description is provided below in conjunction with fig. 3 and 4. Fig. 3 illustrates an effect contrast diagram of an example image anonymization method according to embodiments of the present disclosure. Fig. 4 illustrates an effect contrast diagram of an image anonymization method according to another example of embodiments of the present disclosure.

As shown in fig. 3A, 4 faces in fig. 3A need to be anonymized, that is, 4 regions to be anonymized are provided, and the image anonymizing method according to the embodiment of the present disclosure may be implemented on the 4 regions to be anonymized respectively. In this example, a gaussian blur is described as an example, that is, a multi-layer gaussian blur process is applied to fig. 3A. Specifically, 4 layers (only 4 layers at the male face region are schematically shown in fig. 3B) with sequentially increasing areas may be determined for each face region, and gaussian blur with sequentially decreasing blur radii may be applied to the 4 layers, respectively, to obtain fig. 3B after anonymization processing. As shown in fig. 3B, the 4 face regions all achieve irreversible anonymization, the face details cannot be recognized, and the edges of the face regions are smooth, and no additional edge features exist, so that the processed image is not affected to be used for other purposes such as machine learning training of CV algorithm.

As shown in fig. 4A, 1 license plate of an automobile in fig. 4A needs to be anonymized, that is, has 1 region to be anonymized, and the method for anonymizing images according to the embodiment of the present disclosure may be implemented on the region to be anonymized. In this example, a gaussian blur is described as an example, that is, a multi-layer gaussian blur process is applied to fig. 4A. Specifically, 4 layers with sequentially increased areas can be determined for the license plate region, and Gaussian blur with sequentially decreasing blur radii is respectively applied to the 4 layers, so that a anonymized figure 4B is obtained. As shown in fig. 4B, the license plate is irreversibly anonymized, the license plate number is completely unrecognizable, and the edges of the license plate region are smooth, and no additional edge features exist, so that the processed image is not affected to be used for other purposes such as machine learning training of CV algorithm.

By determining the plurality of layers for anonymizing, sequentially increasing the areas of the plurality of layers from the center to the periphery of the region to be anonymized, sequentially decreasing the anonymizing parameters of the plurality of layers from the center to the periphery of the region to be anonymized, and ensuring that a step transition region exists between different layers while realizing irreversible anonymization of the region to be anonymized, thereby avoiding edge features at the edges of different layers. In addition, the image anonymizing method according to the embodiment of the disclosure can be better compatible with anonymizing features of different categories, for example, better anonymization can be realized on a face area and a license plate area in an image at the same time.

An image anonymizing apparatus according to an embodiment of the present disclosure is described below with reference to fig. 5. Fig. 5 shows a schematic configuration diagram of an image anonymizing apparatus according to an embodiment of the present disclosure. As shown in fig. 5, the image anonymizing apparatus 500 includes a region determining unit 510, a layer determining unit 520, and a processing unit 530. The image anonymizing apparatus 500 may include other units in addition to the 3 units, but since they are irrelevant to the contents of the embodiment of the present invention, a description of these units is omitted here. In addition, since the functions of the image anonymizing apparatus 500 are similar to those of the image anonymizing method 100 described with reference to fig. 1, a repetitive description of some of the same contents is omitted herein.

The region determination unit 510 is configured to determine a region to be anonymized of the image. In the embodiment of the disclosure, an image refers to an image acquired by a camera or a picture frame in a video, and a region to be anonymized of the image refers to a region to be anonymized of the image. For example, if anonymization processing is desired for a face appearing in an image, the region to be anonymized refers to a face region in the image. For another example, if it is desired to anonymize a license plate of an automobile appearing in an image, the region to be anonymized refers to a license plate region in the image. It should be noted that, although the face and the license plate are taken as examples of the region to be anonymized above, the embodiment of the present disclosure is not limited thereto, and the region to be anonymized in the embodiment of the present disclosure may be any image region that needs to be anonymized.

The layer determining unit 520 is configured to determine a plurality of layers to be anonymized according to the size of the region to be anonymized. Specifically, in the embodiment of the present disclosure, multiple layers may be determined in an image according to the size of a region to be anonymized, and anonymization processing may be performed for each layer in the multiple layers, so as to achieve a better anonymization effect. In the embodiment of the present disclosure, the plurality of layers may be two layers, three layers, or more layers, and the number of layers is not particularly limited in the embodiment of the present disclosure.

According to an example of the embodiment of the present disclosure, the greater the proportion of the region to be anonymized in the image, the greater the number of layers to be anonymized may be determined by the layer determining unit 520. In general, the larger the proportion of the region to be anonymized in the image, the higher the resolution of the region to be anonymized or the more image details are contained, so that stronger anonymization processing is required to realize irreversible anonymization. For example, if the tail of an automobile is photographed at a short distance, so that the license plate area in the obtained image occupies a relatively large area, the resolution of the license plate number in the image will be very high, in which case, in order to achieve irreversible anonymization of the license plate, a stronger anonymization process needs to be performed on the license plate. At this time, if the anonymization is performed by using the conventional single-layer gaussian blur method, a large gaussian blur radius is required or the same region is consecutively gaussian blurred a plurality of times, which results in very significant edge features at the edge of the license plate. In contrast, according to the image anonymizing method of the embodiment of the present disclosure, by determining a plurality of layers based on the region to be anonymized and anonymizing the images layer by layer, irreversible anonymization can be achieved without introducing edge features, as will be described in further detail below.

After determining the plurality of layers to be anonymized, the layer determining unit 520 is further configured to determine an area and anonymization parameters of each of the plurality of layers. In examples of embodiments of the present disclosure, each of the plurality of layers may be a square area, a circular area, a polygonal area, etc., to which embodiments of the present disclosure are not particularly limited. In an example of an embodiment of the present disclosure, the anonymization parameter is a parameter employed when anonymizing, and its size reflects the strength of the anonymization. For example, in the case where the image anonymization method according to the embodiment of the present disclosure is based on gaussian blur, the anonymization parameter refers to a gaussian blur radius or a so-called gaussian kernel, and this value represents, for any pixel to be processed, how many surrounding values of pixels are employed for smoothing processing, and the larger the blur radius, the more blurred the processed image.

According to an example of an embodiment of the present disclosure, the layer determining unit 520 may sequentially increment the area of each layer of the plurality of layers from the center to the periphery of the region to be anonymized, and sequentially decrement the anonymization parameter of each layer of the plurality of layers from the center to the periphery of the region to be anonymized. That is, the layer closest to the center of the region to be anonymized has the smallest area and the anonymization parameter is the largest. Taking fig. 2 as an example for illustration, fig. 2 shows an example of multiple layers according to an example of an embodiment of the present disclosure. In fig. 2, the hatched portion indicates the region to be anonymized, and the square region from the center to the periphery of the region to be anonymized is layer 1, layer 2, layer 3, and layer 4 in this order. It can be seen that from layer 1 to layer 4, the area of the layers increases gradually, but the anonymization parameters applied to layer 1 to layer 4 may decrease gradually. Taking anonymization parameter as gaussian blur radius as an example, the gaussian blur radii of layer 1 to layer 4 may be respectively: 15. 11, 7, 5, namely 15 pixels, 11 pixels, 7 pixels and 5 pixels, respectively.

In an example of an embodiment of the present disclosure, the layer determining unit 520 may first determine the area of the first layer and the anonymization parameter based on the size of the region to be anonymized. For example, the area of the first layer may be made smaller than or equal to the area of the region to be anonymized. For example, as shown in fig. 2, in the case where the area to be anonymized has been determined, it may be determined that the area of the first layer is equal to the area of the area to be anonymized, that is, it is determined that the first layer is layer 1, and anonymization parameters of the first layer may be determined, for example, it is determined that the anonymization parameters of layer 1 are 15. Then, the layer determining unit 520 determines the area and anonymization parameters of the remaining layers of the plurality of layers according to the area and anonymization parameters of the first layer. Specifically, as described above, the areas of the remaining layers may be sequentially increased compared to the area of the first layer, for example, the remaining layers may be determined to be layer 2, layer 3, and layer 4, which are sequentially increased in area; and the anonymization parameters of the rest layers are sequentially decreased compared with the anonymization parameters of the first layer, for example, the gaussian blur radii of the layers 2 to 4 may be respectively: 11. 7, 5.

Further, according to an example of an embodiment of the present disclosure, the layer determining unit 520 may make the side length or radius of each layer of the plurality of layers larger than the side length of the previous layer by a predetermined number of pixels. For example, in the case where each layer is a square area, the side length of each layer may be made larger than the side length of the previous layer by a predetermined number of pixels. For another example, in the case where each layer is a circular area, the radius of each layer may be made a predetermined number of pixels larger than the radius of the previous layer. The predetermined number of pixels may be, for example, 4 to 10 pixels, but the embodiment of the present disclosure is not limited thereto, and the specific value of the predetermined number may be determined according to actual needs. For example, in fig. 2, the side length of layer 2 may be made 10 pixels longer than the side length of layer 1, the side length of layer 3 may be made 10 pixels longer than the side length of layer 1, and the side length of layer 4 may be made 10 pixels longer than the side length of layer 1.

Alternatively, according to an example of an embodiment of the present disclosure, the layer determining unit 520 may make a side length difference or a radius difference of an adjacent layer among the plurality of layers equal to a predetermined multiple of a difference of anonymization parameters of the adjacent layer. For example, in the case where each layer is a square region, the difference in side length of adjacent layers among the plurality of layers may be made equal to a predetermined multiple of the difference in anonymization parameters of the adjacent layers. For another example, in the case where each layer is a circular area, the difference in radius of adjacent layers among the plurality of layers may be made equal to a predetermined multiple of the difference in anonymization parameters of the adjacent layers. In the example of the embodiment of the present disclosure, the predetermined multiple may be, for example, 2 times, but the embodiment of the present disclosure is not limited thereto, and a specific value of the predetermined multiple may be determined according to actual requirements.

In this example, the layer determining unit 520 may first determine the area of the first layer and anonymization parameters of the respective layers, and then determine the areas of the remaining respective layers based on the area of the first layer and anonymization parameters of the respective layers. Still taking fig. 2 as an example, assuming that the side length of the square layer 1 is 50 pixels, the gaussian blur radii of the layers 1 to 4 are respectively: 15. 11, 7, 5, the side length of layer 2 can be determined to be 58 pixels by making the side length difference of layer 2 and layer 1 2 be 2 times the blur radius difference of layer 2 and layer 1, and the area of layer 2 is larger than layer 1; similarly, the side lengths of layer 3 and layer 4 may be determined to be 66 pixels and 70 pixels, respectively.

The processing unit 530 is configured to anonymize the image based on the area of each layer and anonymization parameters for each layer of the plurality of layers. For example, in the example of fig. 2, assume that the gaussian blur radii of layers 1 to 4 are respectively: 15. 11, 7, 5, then the layer 1 may be sequentially subjected to a gaussian blur with a blur radius of 15 pixels, the layer 2 may be subjected to a gaussian blur with a blur radius of 11 pixels, the layer 3 may be subjected to a gaussian blur with a blur radius of 7 pixels, and the layer 4 may be subjected to a gaussian blur with a blur radius of 5 pixels.

By means of the image anonymizing device, the multiple layers for anonymizing are determined, the areas of the multiple layers are sequentially increased from the center to the periphery of the region to be anonymized, anonymizing parameters of the multiple layers are sequentially decreased from the center to the periphery of the region to be anonymized, irreversible anonymization of the region to be anonymized can be achieved, meanwhile, a step transition region is ensured to exist between different layers, and therefore edge features at edges of different layers are avoided. In addition, the image anonymizing device according to the embodiment of the disclosure can be better compatible with anonymizing features of different categories, for example, better anonymization can be realized on a face area and a license plate area in an image at the same time.

Furthermore, devices (e.g., image anonymizing devices, etc.) according to embodiments of the present disclosure may also be implemented by way of the architecture of the exemplary computing device shown in fig. 6. Fig. 6 illustrates a schematic diagram of an architecture of an exemplary computing device, according to an embodiment of the present disclosure. As shown in fig. 6, computing device 600 may include a bus 610, one or more CPUs 620, a Read Only Memory (ROM) 630, a Random Access Memory (RAM) 640, a communication port 650 connected to a network, an input/output component 660, a hard disk 670, and the like. A storage device, such as ROM 630 or hard disk 670, in computing device 600 may store various data or files for computer processing and/or communication and program instructions for execution by the CPU. Computing device 600 may also include a user interface 680. Of course, the architecture shown in FIG. 6 is merely exemplary, and one or more components of the computing device shown in FIG. 6 may be omitted as may be practical in implementing different devices. The apparatus according to the embodiments of the present disclosure may be configured to perform the image anonymizing method according to the above-described embodiments of the present disclosure, or to implement the image anonymizing device according to the above-described embodiments of the present disclosure.

Embodiments of the present disclosure may also be implemented as a computer-readable storage medium. Computer readable storage media according to embodiments of the present disclosure have computer readable instructions stored thereon. The image anonymization method according to the embodiments of the present disclosure described with reference to the above figures may be performed when the computer readable instructions are executed by the processor. Computer-readable storage media include, but are not limited to, volatile memory and/or nonvolatile memory, for example. Volatile memory can include, for example, random Access Memory (RAM) and/or cache memory (cache) and the like. The non-volatile memory may include, for example, read Only Memory (ROM), hard disk, flash memory, and the like.

According to an embodiment of the present disclosure, there is also provided a computer program product or a computer program comprising computer readable instructions stored in a computer readable storage medium. The processor of the computer device may read the computer-readable instructions from the computer-readable storage medium, and execute the computer-readable instructions, so that the computer device performs the image anonymization method described in the above embodiments.

Those skilled in the art will appreciate that various modifications and improvements can be made to the disclosure. For example, the various devices or components described above may be implemented in hardware, or may be implemented in software, firmware, or a combination of some or all of the three.

Furthermore, as shown in the present disclosure and claims, unless the context clearly indicates otherwise, the words "a," "an," "the," and/or "the" are not specific to the singular, but may include the plural. The terms "first," "second," and the like, as used in this disclosure, do not denote any order, quantity, or importance, but rather are used to distinguish one element from another. Likewise, the word "comprising" or "comprises", and the like, means that elements or items preceding the word are included in the element or item listed after the word and equivalents thereof, but does not exclude other elements or items. The terms "connected" or "connected," and the like, are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect.

Further, a flowchart is used in this disclosure to describe the operations performed by the system according to embodiments of the present disclosure. It should be understood that the preceding or following operations are not necessarily performed in order precisely. Rather, the various steps may be processed in reverse order or simultaneously. Also, other operations may be added to the processes or a step or steps may be removed from the processes.

Unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this disclosure belongs. It will be further understood that terms, such as those defined in commonly used dictionaries, should be interpreted as having a meaning that is consistent with their meaning in the context of the relevant art and will not be interpreted in an idealized or overly formal sense unless expressly so defined herein.

While the present disclosure has been described in detail above, it will be apparent to those skilled in the art that the present disclosure is not limited to the embodiments described in the present specification. The present disclosure may be embodied as modifications and variations without departing from the spirit and scope of the disclosure, which is defined by the appended claims. Accordingly, the description herein is for the purpose of illustration and is not intended to be in any limiting sense with respect to the present disclosure.

Claims

1. An image anonymizing method, comprising:

determining a region to be anonymized of the image;

determining a plurality of layers to be anonymized according to the size of the region to be anonymized;

determining an area and anonymization parameters of each of the plurality of layers; and

For each layer of the plurality of layers, anonymizing the image based on an area of the layer and anonymizing parameters.

2. The image anonymizing method according to claim 1, wherein determining a plurality of layers to be anonymized according to the size of the region to be anonymized comprises:

the larger the proportion of the region to be anonymized in the image is, the larger the number of layers to be anonymized is determined.

3. The image anonymization method of claim 1, wherein determining an area and anonymization parameters of each of the plurality of layers comprises:

sequentially increasing the area of each layer in the plurality of layers from the center to the periphery of the region to be anonymized; and

and sequentially decreasing anonymization parameters of each layer in the plurality of layers from the center to the periphery of the region to be anonymized.

4. A method of image anonymization according to claim 3, wherein determining the area and anonymization parameter of each of the plurality of layers comprises:

the side length or radius of each layer of the plurality of layers is made a predetermined number of pixels greater than the side length or radius of the previous layer.

5. An image anonymizing method according to claim 4, wherein the predetermined number of pixels is 4 to 10 pixels.

6. A method of image anonymization according to claim 3, wherein determining the area and anonymization parameter of each of the plurality of layers comprises:

the edge length difference or the radius difference of adjacent layers in the plurality of layers is made equal to a predetermined multiple of the difference of anonymization parameters of the adjacent layers.

7. An image anonymizing method according to claim 6, wherein the predetermined multiple is 2 times.

8. The image anonymization method of claim 1, wherein determining an area and anonymization parameters of each of the plurality of layers comprises:

determining the area of the first layer and anonymization parameters based on the size of the region to be anonymized;

and determining the area and anonymization parameters of the rest layers in the layers according to the area and anonymization parameters of the first layer.

9. The image anonymizing method according to claim 8, wherein the area of the first layer is less than or equal to the area of the region to be anonymized, and wherein determining the area and anonymization parameters of the remaining layers of the plurality of layers according to the area and anonymization parameters of the first layer comprises:

Sequentially increasing the area of the rest of the plurality of layers compared to the area of the first layer; and is also provided with

Anonymization parameters of the remaining layers of the plurality of layers are sequentially decremented as compared to anonymization parameters of the first layer.

10. An image anonymizing method according to claim 1, wherein the anonymizing process is gaussian blur, and the anonymizing parameter is a gaussian blur radius.

11. An image anonymizing apparatus, the apparatus comprising:

a region determining unit configured to determine a region to be anonymized of the image;

a layer determining unit configured to determine a plurality of layers to be anonymized according to the size of the region to be anonymized, and determine an area and anonymization parameters of each layer in the plurality of layers; and

and a processing unit configured to anonymize the image based on an area of the layer and anonymization parameters for each layer of the plurality of layers.

12. The image anonymizing apparatus according to claim 11, wherein the area determining unit is further configured to:

the larger the area to be anonymized is, the larger the number of layers to be anonymized is determined.

13. An image anonymizing apparatus comprising:

one or more processors; and

one or more memories having stored therein computer readable instructions that, when executed by the one or more processors, cause the one or more processors to perform the method of any of claims 1-10.

14. A computer readable storage medium having stored thereon computer readable instructions which, when executed by a processor, cause the processor to perform the method of any of claims 1-10.

15. A computer program product comprising computer readable instructions which, when executed by a processor, cause the processor to perform the method of any of claims 1-10.