WO2021098572A1

WO2021098572A1 - Image processing method, apparatus and device, and computer-readable storage medium

Info

Publication number: WO2021098572A1
Application number: PCT/CN2020/128189
Authority: WO
Inventors: 周扬; 刘杰
Original assignee: Oppo广东移动通信有限公司
Priority date: 2019-11-20
Filing date: 2020-11-11
Publication date: 2021-05-27
Also published as: CN114730360A

Abstract

An image processing method, apparatus and device, and a computer-readable storage medium. The method comprises: performing, according to a preset object category, target detection on an image to be processed to obtain at least one piece of object region information corresponding to at least one detection object (S101); screening the at least one detection object according to the at least one piece of object region information, and determining a target detection object from the at least one detection object, wherein object region information of the target detection object conforms to preset standard region information required by image processing (S102); and performing image processing on the target detection object (S103).

Description

Image processing method, device, equipment and computer readable storage medium

Technical field

This application relates to image processing technology in the field of artificial intelligence, and in particular to an image processing method, device, device, and computer-readable storage medium.

Background technique

Face detection has always been one of the important topics in computer vision research, and it plays an important role in daily applications such as face unlocking and video surveillance.

In the face unlocking scene, the face image contained in the image can be detected by the face detection technology, and then face verification and face unlocking are performed based on the detected face image. However, since the images actually captured by the image acquisition device may randomly contain images of background faces that do not belong to the target person, the current face detection methods will detect both the target face and the background face as the face detection results. , And performing face unlock on the face detection result that contains the background face will cause the face unlock failure, thereby reducing the accuracy of target detection and image processing.

Summary of the invention

The embodiments of the present application provide an image processing method, device, and computer-readable storage medium, which can improve the accuracy of target detection and image processing.

The technical solutions of the embodiments of the present application are implemented as follows:

An embodiment of the application provides an image processing method, including:

Performing target detection on the image to be processed according to the preset object category, to obtain at least one object region information corresponding to the at least one detection object;

According to the at least one object area information, the at least one detection object is screened, and the target detection object is determined from the at least one detection object; the object area information of the target detection object meets the preset standard required by image processing Area information

Image processing is performed on the target detection object.

An embodiment of the application provides an image processing device, including:

The target detection module is configured to perform target detection according to the to-be-processed image of the preset object category to obtain at least one object area of at least one detection object of the preset object category;

A screening module, configured to screen the at least one detection object according to the at least one object area, and determine a target detection object from the at least one detection object;

The determining module is used to perform image processing on the target detection object.

Memory, used to store executable instructions;

The processor is configured to implement the image processing method provided in the embodiment of the present application when executing the executable instructions stored in the memory.

The embodiment of the present application provides a computer-readable storage medium that stores executable instructions for causing a processor to execute to implement the image processing method provided by the embodiment of the present application.

The embodiments of this application have the following beneficial effects:

When at least one detection object is obtained through target detection, at least one object area information corresponding to the at least one detection object can be used to exclude detection objects that do not meet the preset standard area information from the at least one detection object, so that the final target detection object is determined It can be matched with the requirements of image processing, thereby improving the accuracy of target detection, and further improving the accuracy of image processing based on target detection.

Description of the drawings

FIG. 1 is an optional structural schematic diagram of the image processing system architecture provided by an embodiment of the present application;

Fig. 2 is a schematic diagram of a face unlocking process provided by an embodiment of the present application;

FIG. 3 is a schematic diagram of an optional structure of an image processing apparatus provided by an embodiment of the present application;

Fig. 4 is an optional flowchart of the image processing method provided by an embodiment of the present application.

Fig. 5 is a schematic diagram of a face detection result provided by an embodiment of the present application.

FIG. 6 is a schematic diagram of an optional flow chart of the target detection process provided by an embodiment of the present application.

Fig. 7 is an optional flowchart of the image processing method provided by an embodiment of the present application.

FIG. 8 is a schematic diagram of an object area detected from an image to be processed according to an embodiment of the present application.

FIG. 9 is a schematic diagram of a process of filtering out one of the two bounding boxes according to an embodiment of the present application.

FIG. 10 is an optional flowchart of an image processing method provided by an embodiment of the present application.

FIG. 11 is an optional flowchart of an image processing method provided by an embodiment of the present application.

FIG. 12 is an optional flowchart of an image processing method provided by an embodiment of the present application.

FIG. 13 is an optional flowchart of an image processing method provided by an embodiment of the present application.

FIG. 14 is an optional flowchart of an image processing method provided by an embodiment of the present application;

FIG. 15 is a schematic diagram of a flow of unlocking a face according to face input provided by an embodiment of the present application;

FIG. 16 is an optional flowchart of an image processing method provided by an embodiment of the present application.

Detailed ways

In order to make the purpose, technical solutions, and advantages of this application clearer, the application will be further described in detail below in conjunction with the accompanying drawings. The described embodiments should not be regarded as limiting the application. Those of ordinary skill in the art have not made any suggestions. All other embodiments obtained under the premise of creative labor belong to the scope of protection of this application.

In the following description, “some embodiments” are referred to, which describe a subset of all possible embodiments, but it is understood that “some embodiments” may be the same subset or different subsets of all possible embodiments, and Can be combined with each other without conflict.

In the following description, the term "first\second\third" is only used to distinguish similar objects, and does not represent a specific order of objects. Understandably, "first\second\third" Where permitted, the specific order or sequence can be interchanged, so that the embodiments of the present application described herein can be implemented in a sequence other than those illustrated or described herein.

Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the technical field of this application. The terminology used herein is only for the purpose of describing the embodiments of the application, and is not intended to limit the application.

Before describing the embodiments of the present application in further detail, the terms and terms involved in the embodiments of the present application will be described. The terms and terms involved in the embodiments of the present application are applicable to the following interpretations.

1) Target detection: Image classification, target detection, and image segmentation are three major tasks in the field of computer vision. Among them, the image classification task is concerned with the whole, while the target detection is concerned with the specific object target, and it is required to obtain the category information and position information of this target at the same time. Target detection is to isolate the target object of interest from the background as the target object through the recognition and analysis of the foreground and background of the picture, and output the confidence, position and size information of the target object as the boundary information of the target object; among them, the position The sum size is usually expressed by the coordinates of the rectangular bounding box.

2) Fast Single Shot MultiBox Detector (SSD): SSD is a target detector based on a neural network model, which can be applied to multiple target object categories. A key feature of the SSD model is to use multi-scale convolutional bounding boxes connected to multiple feature maps for output at the high level of the network. This network expression method can effectively simulate different bounding box aspect ratios.

3) You Only Look Once (YOLO): Object recognition and localization algorithm based on deep neural network, only one convolutional neural network operation can locate the target object and the location of the target object in the image. YOLO is characterized by its fast running speed and can be used in real-time systems.

Face image detection is also referred to as face detection (Face Detection), which refers to the process of judging whether there is a face image in the input image and determining the specific location of all face image regions. At present, face image detection usually uses target detection based on convolutional networks, which is mainly composed of two major components: a front-end feature extractor and a back-end detector. Among them, the front-end feature extractor is used to extract image features from the image to be processed, and the back-end detector is used to predict the image corresponding to the detection target from the image to be processed based on the image features extracted by the front-end A bounding box is generated around the area where the image is located to calibrate the detection target.

The embodiments of the present application provide an image processing method, device, equipment, and computer-readable storage medium, which can improve the accuracy of target recognition. The following describes exemplary applications of the image processing equipment provided in the embodiments of the present application. The embodiments of the present application provide The devices can be implemented as notebook computers with image capture devices, tablet computers, desktop computers, set-top boxes, mobile devices (for example, mobile phones, portable music players, personal digital assistants, dedicated messaging devices, portable game devices), etc. Type of user terminal. In the following, an exemplary application when the device is implemented as a terminal will be explained.

Referring to FIG. 1, FIG. 1 is a schematic diagram of an optional architecture of an image processing system 100 provided by an embodiment of the present application. In order to support an image processing application, a terminal 400 is connected to a server 200 through a network 300. The network 300 may be a wide area network or a local area network. Or a combination of the two.

The terminal 400 is used to collect the face image of the target person through the image acquisition device, and perform image decoding, face detection, and face verification processing on the face image according to the process shown in FIG. 2, and then according to the verification of the face verification As a result, it is determined whether to unlock the face. In the face detection and face verification process shown in FIG. 2, the terminal 400 is used to set the face category as the preset object category, the decoded image as the image to be processed, and perform target detection on the image to be processed according to the preset object category , Obtain at least one object area information corresponding to at least one detection object; filter at least one detection object according to the at least one object area information, and determine the target detection object from the at least one detection object; the object area information of the target detection object conforms to image processing The required standard area information; image processing is performed on the target detection object; and the image processing result is displayed on the graphical interface 401. The server 200 is used to obtain pre-stored standard face images from the database 500, and provide the standard face images to the terminal through the network 300 when the terminal 400 performs face verification, so that the terminal can complete face verification and face unlocking and other images deal with.

Exemplarily, when the preset object category is a face category, in the face unlocking scene, the terminal 400 may first prompt on the graphical interface 401 that the face is to be unlocked, and the terminal 400 obtains the image to be processed through the image acquisition device, and The image to be processed may be subject to face category target detection, at least one face image is detected as at least one detection object from the image to be processed, and at least one object region information corresponding to the at least one face image is obtained; wherein, at least one object Each object area information in the area information may be the confidence level predicted by the target detection network for the face image, as well as the size and position of the rectangular area occupied by the face image; the terminal 400 may determine at least one object area information according to at least one object area. One face image is screened, the background face image is excluded from at least one face image, and the target face image is determined; the terminal 400 obtains the pre-stored standard face image from the database 500 through the server 200, and according to the standard face image Perform face verification on the target face image, and if the verification passes, it is determined that the face unlock is successful; if the verification fails, it is determined that the face unlock has failed, and a failure message is displayed on the graphical interface 401 of the terminal 400.

In some embodiments, the server 200 may be an independent physical server, a server cluster or a distributed system composed of multiple physical servers, or it may provide cloud services, cloud databases, cloud computing, cloud functions, cloud storage, Cloud servers for basic cloud computing services such as network services, cloud communications, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms. The terminal 400 may be a smart phone, a tablet computer, a notebook computer, a desktop computer, a smart speaker, a smart watch, etc., but is not limited to this. The terminal and the server can be directly or indirectly connected through wired or wireless communication, which is not limited in the embodiment of the present invention.

Referring to FIG. 3, FIG. 3 is a schematic structural diagram of a terminal 400 provided by an embodiment of the present application. The terminal 400 shown in FIG. 3 includes: at least one processor 410, a memory 450, at least one network interface 420, and a user interface 430. The various components in the terminal 400 are coupled together through the bus system 440. It can be understood that the bus system 440 is used to implement connection and communication between these components. In addition to the data bus, the bus system 440 also includes a power bus, a control bus, and a status signal bus. However, for clear description, various buses are marked as the bus system 440 in FIG. 3.

The processor 410 may be an integrated circuit chip with signal processing capabilities, such as a general-purpose processor, a digital signal processor (DSP, Digital Signal Processor), or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware Components, etc., where the general-purpose processor may be a microprocessor or any conventional processor.

The user interface 430 includes one or more output devices 431 that enable the presentation of media content, including one or more speakers and/or one or more visual display screens. The user interface 430 also includes one or more input devices 432, including user interface components that facilitate user input, such as a keyboard, a mouse, a microphone, a touch screen display, a camera, and other input buttons and controls.

The memory 450 may be removable, non-removable, or a combination thereof. Exemplary hardware devices include solid-state memory, hard disk drives, optical disk drives, and so on. The memory 450 optionally includes one or more storage devices that are physically remote from the processor 410.

The memory 450 includes volatile memory or nonvolatile memory, and may also include both volatile and nonvolatile memory. The non-volatile memory may be a read only memory (ROM, Read Only Memory), and the volatile memory may be a random access memory (RAM, Random Access Memory). The memory 450 described in the embodiment of the present application is intended to include any suitable type of memory.

In some embodiments, the memory 450 can store data to support various operations. Examples of these data include programs, modules, and data structures, or a subset or superset thereof, as illustrated below.

Operating system 451, including system programs used to process various basic system services and perform hardware-related tasks, such as framework layer, core library layer, driver layer, etc., used to implement various basic services and process hardware-based tasks;

The network communication module 452 is used to reach other computing devices via one or more (wired or wireless) network interfaces 420. Exemplary network interfaces 420 include: Bluetooth, Wireless Compatibility Authentication (WiFi), and Universal Serial Bus ( USB, Universal Serial Bus), etc.;

The presentation module 453 is used to enable the presentation of information via one or more output devices 431 (for example, a display screen, a speaker, etc.) associated with the user interface 430 (for example, a user interface for operating peripheral devices and displaying content and information) );

The input processing module 454 is configured to detect one or more user inputs or interactions from one of the one or more input devices 432 and translate the detected inputs or interactions.

In some embodiments, the image processing device provided by the embodiments of the present application can be implemented in software. FIG. 3 shows the image processing device 455 stored in the memory 450, which can be software in the form of programs and plug-ins, including the following Software modules: target detection module 4551, screening module 4552, and image processing module 4553. These modules are logical, so they can be combined or further split arbitrarily according to the realized functions.

The function of each module will be explained below.

In other embodiments, the image processing apparatus provided in the embodiments of the present application may be implemented in hardware. As an example, the image processing apparatus provided in the embodiments of the present application may be a processor in the form of a hardware decoding processor, which is programmed To execute the image processing method provided by the embodiments of the present application, for example, a processor in the form of a hardware decoding processor may adopt one or more application specific integrated circuits (ASIC, Application Specific Integrated Circuit), DSP, and programmable logic device (PLD). , Programmable Logic Device, Complex Programmable Logic Device (CPLD, Complex Programmable Logic Device), Field-Programmable Gate Array (FPGA, Field-Programmable Gate Array) or other electronic components.

The image processing method provided in the embodiment of the present application will be described in conjunction with the exemplary application and implementation of the terminal provided in the embodiment of the present application.

Referring to FIG. 4, FIG. 4 is an optional flowchart of the image processing method provided by an embodiment of the present application, which will be described in conjunction with the steps shown in FIG. 4.

S101: Perform target detection on an image to be processed according to a preset object category, and obtain at least one object area information corresponding to at least one detected object.

In the embodiment of the present application, the image processing device may perform target detection on the image to be processed according to the preset object category by using the target detection method, detect at least one detection object belonging to the preset object category from the image to be processed, and obtain at least one correspondingly. At least one object area information corresponding to the detected object.

In S101, the preset object category is a preset specific object target. Exemplarily, in a face unlocking scene, the preset object category may be a human face. In an autonomous driving scene, the preset object category may be a road marking or an obstacle, etc., and the specific selection is made according to the actual situation. The implementation of this application The examples are not limited.

In S101, at least one piece of object area information is a detection result of the image processing apparatus performing target detection on the image to be processed according to a preset object category. The image processing device can analyze the foreground and background of the image to be processed through the target detection method, predict from the background at least one detection object corresponding to the preset object category, and perform boundary calibration on the image area where each detection object is located, and Obtain the object area information of the detection object according to the position of the boundary in the image, the included range, and the confidence of the detection object in the boundary, so as to obtain at least one object area information. Exemplarily, the image processing apparatus may use the size and position of the boundary corresponding to each detection object and the confidence level corresponding to the detection object as at least one object area information.

In S101, the confidence level represents the probability that the detected object inferred by the image processing device through the target detection method belongs to the preset object category.

In some embodiments, the object area information may be visually displayed as a two-dimensional bounding box in the graphical interface of the terminal. Figure 5 shows two bounding boxes, bounding box 1 and bounding box 2, obtained by the image processing device for face detection on the image to be processed when the image to be processed contains two faces, wherein bounding box 1 is based on The object area information corresponding to the detection target face 1 is visualized, and the bounding box 2 is obtained by visualizing the object area information corresponding to the detection target face 2. The bounding box 1 and the bounding box 2 respectively surround the face 1 and the face 2 to display.

In some embodiments, the image processing device can implement target detection of the image to be processed through the SSD network model, or through the YOLO or YOLO2 model, and the specific selection is based on actual conditions, which is not limited in the embodiment of the present application.

In some embodiments, the image processing device uses the YOLO model to perform target detection and obtains at least one object area information as shown in FIG. 6. For a picture with a dog, a bicycle, and a car, the image processing device can divide the picture into For multiple small squares, the feature map 1 corresponding to the picture is obtained, and the image processing device uses the YOLO model to perform target detection on the image content in each small square in the feature map 1 to generate at least one corresponding to each small square Prediction box, finally filter out incorrect prediction boxes from at least one prediction box, and finally determine the bounding box A corresponding to the detection target dog, the bounding box B corresponding to the detection target bicycle, and the bounding box C corresponding to the detection target car, and the image processing device The respective information of the bounding box A, the bounding box B, and the bounding box C, such as size, position, and confidence, are used as at least one object area information.

S102. Filter at least one detection object according to the at least one object area information, and determine the target detection object from the at least one detection object; the object area information of the target detection object meets the preset standard area information required by image processing.

In S102, after the image processing apparatus obtains at least one object area information, it will exclude the object area that does not meet the prior knowledge from the at least one object area information based on the prior knowledge of the target detection object based on the at least one object area information Information to determine the target detection object for further image processing.

In S102, the image in the at least one object region information is at least one detection object of a preset object category predicted by the image processing device from the image to be processed, but not all the at least one detection object requires further image processing. Exemplarily, in the face unlocking scene, the front camera may capture the facial images of other people near the owner. In this way, at least one detection object detected by the image processing device through the target detection method includes the owner's face Images, including facial images of non-hosts. However, for face unlocking, only the face image of the owner is the object that needs to be identified and verified, and other face images identified will cause misjudgment and cause unlocking failure. Therefore, the image processing device needs to screen the identified at least one detection object according to the at least one object area information, and determine the target object area information that meets the image recognition target.

In S102, the image processing device may gradually exclude differences from the preset standard area information from at least one object area information based on the preset standard area information obtained according to the prior knowledge and based on multiple parameter types in the preset standard area information. The corresponding object area information until the finally retained object area information is used as the target object area information.

In some embodiments, in the face unlocking scene, since the image used for face unlocking is usually taken within a range close to the face, such as the arm length range, the image based on the face unlocking Prior knowledge, set preset standard area information. Exemplary: The usual image characteristics of the face unlock image are: the image to be processed contains an image of a target person, and the target person image has a higher confidence level, for example, the confidence level is higher than 0.4; the target person image Located at the center of the image to be processed; the target person image occupies a larger image area, for example, the target person image occupies 0.3 to 0.9 of the total area of the image to be processed. According to the above image characteristics, the preset standard area information may be: the confidence of the detection object should be greater than 0.4, the distance from the area occupied by the detection object to the center of the image to be processed is less than the preset distance threshold; the area occupied by the detection object is at least Should be 0.3 to 0.9 of the total area of the image to be processed.

In S102, the preset standard area information and the at least one object area information are area information containing the same parameter type. In some embodiments, each object area information in the at least one object area information includes one of size, position, and confidence. At least one item, S102 shown in FIG. 4 can be implemented through S1021 to S1022 as shown in FIG. 7, as follows:

S1021, according to the at least one object area information, exclude from the at least one object area object areas whose confidence, size, or position does not meet the preset standard area information, so as to determine the target object area; at least one object area is at least one object area information The range of the area represented.

In S1021, at least one object area is an area range characterized by at least one object area information. In some embodiments, the at least one object area information may be in the form of a data list, each item in the list represents one object area information, and each object area information gives the confidence of the detected object in the form of an array; and the detected object The size of the occupied object area and the position of the center point of the object area. Exemplarily, if the area occupied by each detection object is normalized to a rectangular object area, the information of each object area can be expressed as (cx, cy, ω, h, conf), where conf is the confidence level, ( cx, cy) is the position in the object area information, that is, the coordinates of the center point of the object area, and (ω, h) is the size in the object area information, that is, the width and height of the object area. Exemplarily, as shown in FIG. 8, when the preset object category is a dog, the image processing device performs target detection on the image to be processed 70 to obtain an image in which the detection object is a dog, and the object region information corresponding to the detection object is (cx, cy, ω, h, conf), the object area corresponding to the detection object is the object area 71.

In S1021, according to the parameter type included in the at least one object area information, the image processing apparatus may compare the confidence level of the at least one object area information with the confidence level of the preset standard area information, and exclude the information that is different from the preset standard area information. Object areas with large differences in confidence; the image processing device may also compare the size of at least one object area information with the size of the preset standard area information, and exclude object areas with a large difference from the size of the preset standard area information; The image processing device may also compare the position of the at least one object area information with the position of the preset standard area information, and exclude the object area that is significantly different from the position of the preset standard area information. The image processing device uses the object area finally remaining after the above-mentioned screening as the target object area.

It should be noted that, in the embodiment of the present application, when screening at least one object area in different dimensions, according to actual application needs, different priority orders can be set for the screening methods of different dimensions, or different dimensions can also be selected. One or more of the screening methods are selected and combined to achieve screening of at least one target area, which is specifically selected according to actual conditions, which is not limited in the embodiment of the present application.

It should be noted that in this embodiment of the application, at least one object area information can be directly compared with preset standard area information to determine whether the at least one object area information meets the preset standard information, or according to actual application needs, First, perform a reasonable mathematical transformation on at least one object area information, and then compare it with the preset standard area information. The specific selection is made according to actual conditions, which is not limited in the embodiment of the present application.

S1022, using the detection object in the target object area as the target detection object.

In S1022, the image processing device regards the detection object in the target object area as the target detection object, and performs the next image processing on the target detection object.

In some embodiments, based on the two bounding boxes shown in FIG. 5, the image processing device can filter the two bounding boxes according to their respective size, position, and confidence information, and exclude the non-compliance with the preset standard. The bounding box 2 of the region information, the bounding box 1 is determined from the two bounding boxes as the target bounding box, and the face 1 in the target bounding box is used as the target detection object for the next face unlocking process, such as Shown in Figure 9.

S103: Perform image processing on the target detection object.

In S103, when the image processing apparatus obtains the target detection object, the next image processing can be performed on the target detection object.

In some embodiments, in the face unlocking scene, the target detection object is the target face image, when the image processing device can start the face verification process on the target face image, extract the image features from the target face image, and combine The image features of the target face image are compared with the pre-entered main face image, and it is determined whether it can be successfully unlocked according to the comparison result, thereby completing the face unlocking process.

It can be understood that, in the embodiment of the present application, when at least one detection object is obtained through target detection, the image processing apparatus can exclude at least one detection object from at least one detection object that does not conform to the preset by using at least one object region information corresponding to the at least one detection object The detection object of the standard area information enables the final target detection object to match the requirements of image processing, thereby improving the accuracy of target detection, and further improving the accuracy of image processing based on target detection.

In some embodiments, refer to FIG. 10, which is an optional flowchart of the method provided in the embodiment of the present application. S1021 shown in FIG. 7 can be implemented through S201-S202, which will be described in combination with each step.

S201. In at least one object area, exclude object areas whose confidence is less than a preset confidence threshold, so as to obtain N remaining object areas.

In S201, the preset standard area information includes a preset reliability threshold. When the confidence of the object area is less than the preset confidence threshold, it means that the detection object in the object area has a low probability of belonging to the preset object category, that is, the detection object may not be the target to be processed in the image processing process. Therefore, the image processing apparatus excludes the object areas whose confidence is less than the preset confidence threshold in at least one object area, so as to obtain N remaining bounding boxes.

In some embodiments, the preset reliability threshold may be 0.4.

S202. When N is greater than the preset number threshold, exclude from the N remaining object areas the object areas whose size or position does not meet the preset standard area information, so as to determine the target object area; the preset number threshold is a non-zero positive integer .

In S202, when N is greater than the preset number threshold, the image processing device may continue to filter the N remaining object areas, and compare the object area information of the N remaining object areas with the preset standard area information from the dimensions of size or position. Contrast, exclude the object area whose size or position does not meet the expectations, until the target object area is finally determined.

In S202, the preset number threshold is a non-zero positive integer. Exemplarily, the preset number threshold may be 1. When N is greater than 1, it means that there are still multiple object regions in the image to be processed after confidence filtering, and the image processing device needs to filter again until the final result is determined. Target area.

In S202, the size of each remaining object area may be the width and height of the remaining object area. The image processing device may perform screening again according to the width and height of the N remaining bounding boxes, and further exclude bounding boxes that do not meet the preset standard area information in the size dimension.

In some embodiments, the preset standard area information may include a standard aspect ratio, and the image processing device may obtain the aspect ratio of each remaining object area according to the width and height of each remaining object area in the N remaining object areas , Which can then exclude the remaining object areas whose aspect ratio and standard aspect ratio difference exceeds the preset aspect ratio range, and retain the remaining objects whose aspect ratio and standard aspect ratio difference is within the preset difference range area.

In some embodiments, the preset standard area information includes a preset area threshold, and the image processing device may also obtain the area of each remaining object area according to the width and height of each remaining object area, and then may exclude the area less than the preset area from it. Set the remaining object area with an area threshold, and reserve the remaining object area with an area greater than the preset area threshold. . The specific screening method can be selected according to the actual situation, which is not limited in the embodiment of the present application.

It is understandable that, in this embodiment of the application, the image processing device may first exclude the object areas whose confidence is lower than the preset confidence threshold from the at least one object area information based on the confidence of the at least one object area information, thereby reducing The object area with too low confidence interferes with the image processing process, and the accuracy of image processing is improved; in addition, the image processing device can further filter the N remaining object areas filtered based on the confidence level until the target is obtained. The bounding box thus further improves the accuracy of the obtained target object area, and further improves the accuracy of image processing based on the target object area.

In some embodiments, refer to FIG. 11, which is an optional flowchart diagram of the method provided in an embodiment of the present application. S202 shown in FIG. 10 can be implemented through S2021-S2023, which will be described in combination with each step.

S2021. Calculate the area of each remaining object area in the N remaining object areas according to the size of the N remaining object areas, so as to determine the first object area with the first large area and the second object area with the second large area among the N remaining objects. 2. Object area.

In S2021, the image processing device may calculate the area of the N remaining object regions based on the size of the N remaining object regions, for example, the width and height of the remaining object regions.

In S2021, the image processing device may sort the N remaining object areas according to the areas of the N remaining object areas, so as to determine the first object area with the largest area and the second object area with the second largest area.

S2022, exclude an object area with an area smaller than the second object area from the N remaining object areas.

In S2022, the image processing device may exclude the object area whose area is smaller than the second object area from the N remaining object areas, and only retain the first object area and the second object area, and then remove the object area from the first object area and the second object area. Determine the bounding box of the target.

S2023: Exclude object areas whose areas or positions do not meet preset standard area information from the first object area and the second object area, so as to determine the target object area.

In S2023, after the image processing device excludes the object area whose area is smaller than the second object area, it continues to filter based on the reserved area and position of the first object area and the second object area until the target object area is determined.

In some embodiments, refer to FIG. 12, which is an optional flowchart of the method provided by the embodiment of the present application. S2023 shown in FIG. 11 can be implemented through S301-S303, which will be described in combination with each step.

S301: When the area of the second object area is less than a preset area threshold, determine the first object area as the target object area.

In S301, the preset standard area information includes a preset area threshold. When the area in the second object area is less than the preset area threshold, it means that the area of the second object area is too small and the possibility of becoming the target object area is low. The image processing device excludes the second object area and determines the first object area as Target area.

In some embodiments, the preset area threshold may be determined according to the preset minimum area ratio of the image to be processed. Exemplarily, if the size of the image to be processed is 640*400, the preset minimum area ratio is 25%. Therefore, the image processing device may set the preset area threshold to 640*400*25%, that is, 640.

S302: When the area of the first object area and the area of the second object area are both greater than the preset area threshold, determine whether the area ratio of the second object area to the first object area is greater than the preset proportion threshold.

In S302, the preset standard area information includes a preset proportion threshold. When the areas of the first object area and the second object area are both greater than the preset area threshold, the image processing device may compare the areas of the first object area and the second object area with each other, based on the area ratio and the preset proportion threshold. Continue to filter until the target area is determined.

S303: When the area ratio is greater than the preset proportion threshold, exclude the object areas whose positions do not meet the preset standard area information from the first object area and the second object area, thereby determining the target object area.

In S303, when the area ratio of the second object area to the first object area is greater than the preset proportion threshold, it means that the area of the first object area is not much different from the area of the second object area, and the image processing device can be based on the first object area. Re-screening is performed with the position of the second object area, and the object area whose position does not meet the preset standard area information is excluded from the first object area and the second object area, thereby determining the target object area.

In some embodiments, the preset proportion threshold may be 0.36.

In some embodiments, after S302, S304 may be executed, which will be described in combination with each step.

S304: When the area ratio is less than the preset proportion threshold, the first object area is determined as the target object area.

In S304, when the area ratio is less than the preset proportion threshold, it means that the area of the second object area is much smaller than the first object area, and the first object area is more likely to be the target object area. Therefore, the image processing device can exclude the second object area and determine the first object area as the target object area.

In some embodiments, after S2021, S2024 may be executed as follows:

S2024: When the area of each remaining object area is less than the preset area threshold, end the image processing flow, and prompt that no valid target is detected.

In S2024, when the area of each remaining object area is less than the preset area threshold, it means that the area of each remaining object area identified by the target detection is too small to perform further image processing based on any one of them, so the image The processing device can end the image processing flow, prompting that no valid target has been detected.

It is understandable that, in this embodiment of the application, the image processing device can filter the N remaining object areas based on the area information corresponding to the size, and exclude the object areas that are too small, leaving only the first and second largest areas. There are two object areas, thereby reducing the interference of small object areas on the image processing process and improving the accuracy of image processing. In addition, the image processing device may further filter the first large object area and the second large object area based on their positions, and finally determine the target object area, thereby further ensuring the accuracy of image processing.

In some embodiments, referring to FIG. 13, FIG. 13 is an optional flowchart of the method provided in an embodiment of the present application. S303 shown in FIG. 12 can be implemented through S3031-S3033, which will be described in combination with each step.

S3031, according to the location of the first object area and the location of the second object area, respectively calculate the first distance from the first object area to the image center of the image to be processed and the second distance from the second object area to the image center.

In S3031, the position of the first object area is the coordinates of the center point of the first object area, the position of the second object area is the coordinates of the center point of the second object area, and the image processing device calculates the coordinates of the center point of the first object area. The first distance is obtained by calculating the distance from the coordinates to the center of the image to be processed, and the image processing device calculates the distance between the coordinates of the center point of the second object area and the coordinates of the center point of the image to be processed to obtain the second distance.

In some embodiments, the process of calculating the first distance and the second distance by the image processing apparatus may be implemented through S401-S404, which will be described in conjunction with each step.

S401: Determine the first abscissa of the first object area according to the location of the first object area; the location of the first object area is the coordinate of the center point of the first object area.

In S401, the location of the first object area is the center point coordinates of the first object area on the terminal screen, and the image processing device may determine the first abscissa corresponding to the first object area through the center point coordinates of the first object area .

S402: Determine the second abscissa corresponding to the second object area according to the location of the second object area.

In S402, the location of the second object area is the center point coordinates of the second object area on the terminal screen, and the image processing device may determine the second abscissa corresponding to the second object area through the center point coordinates of the second object area .

S403: Calculate the first lateral distance between the first abscissa and the abscissa of the image center point of the image to be processed, and use the ratio of the first lateral distance to the width of the first object area as the first distance.

S404: Calculate the second intermediate distance between the second abscissa and the vertical center line, and use the ratio of the second intermediate distance to the width of the second object area as the second distance.

In S403, the image processing device calculates the difference between the first abscissa and the abscissa of the image center point as the first lateral distance; in S404, the image processing device calculates the difference between the first abscissa and the abscissa of the image center point. The difference between, as the second lateral distance. Since the size of the first object area and the second object area may be different, in order to reduce the influence of the size of the object area on the distance calculation, the image processing device may normalize the first lateral distance and the second lateral distance, and The first lateral distance is divided by the width of the first object area as the first distance; the second lateral distance is divided by the width of the second object area as the second distance.

In some embodiments, when the abscissa of the center point of the first object area is x and the size of the image to be processed is 640*400, the abscissa of the image center point of the image to be processed is 320. The image processing device may use the absolute value of x-320 as the first distance.

S3032. When any one of the first distance and the second distance is greater than the preset distance threshold, use an object area that is less than the preset distance threshold from the center of the image as the target object area.

In S3032, the preset standard area information includes a preset distance threshold. When any one of the first distance and the second distance is greater than the preset distance threshold, it means that the object area corresponding to the distance is farther from the image center and is less likely to be the target object area, and the image processing device will be less than the preset distance The object area corresponding to the threshold distance is regarded as the target object area.

In some embodiments, the preset distance threshold can be flexibly set according to the width of the first object area and the second object area. Illustratively, for the first object area, the preset distance threshold is set to be less than the width of the first object area. 1.5 times, for the second object area, the preset distance threshold is set to 1.5 times the width of the second object area. In this way, the first object area and the second object area of different widths can be compared according to the preset distance thresholds corresponding to the respective widths.

In some embodiments, when the first distance is |x-640/2| and the width of the first object area is w, if |x-640/2| is less than 1.5w, the image processing device may determine that the first distance is greater than Corresponding to the preset distance threshold, the first object area is far from the center of the image, so that the first object area is excluded from the first object area and the second object area.

S3033: When both the first distance and the second distance are less than the preset distance threshold, determine that the object area corresponding to the smallest distance in the first distance and the second distance is the target object area.

In S3033, when the first distance and the second distance are both less than the preset distance threshold, it indicates that the distances between the first object area and the second object area and the image center point are all within a reasonable range, and the image processing device may further The one distance is compared with the second distance, and the smallest distance between the first distance and the second distance, that is, the object area closer to the center point of the image is taken as the object area.

In some embodiments, S3034 may be included after S3031, as follows:

S3034: When both the first distance and the second distance are greater than the preset distance threshold, the image processing procedure is ended, and a prompt is not detected that a valid target is detected.

In the embodiment of the present application, when the first distance and the second distance are both greater than the preset distance threshold, it means that the first object area and the second object area are both far from the center of the image to be processed, and the image processing device ends the image processing flow and prompts No valid target was detected.

It is understandable that, in the embodiment of the present application, the image processing device can filter at least one object region information layer by layer through the three dimensions of confidence, size, and position, and finally retain high confidence, large area, and high reliability. The object area closer to the center of the image to be processed is used as the target object area, thereby improving the accuracy of locating the target object area, and further improving the accuracy of image processing based on the target object area.

In some embodiments, refer to FIG. 14, which is an optional flowchart of the method provided by an embodiment of the present application. When the preset object category is a face category, the target detection object is the target face, as shown in FIG. 4 The out S103 can be implemented through S1031-S1032, which will be described in combination with each step.

S1031. Perform face verification on the target face according to the pre-entered standard face, and obtain a verification result.

In S1031, when the preset object category is a face category, at least one detection object is at least one face, and the image processing device can determine the target detection object from the at least one face through the above-mentioned S101-S102 process, that is, the target person face. The image processing device can perform image matching such as face comparison according to the pre-entered standard face and the target face, so as to realize the face verification process of the recognition target, and obtain the verification result of the face verification according to the matching result.

In some embodiments, when the matching degree between the recognition target and the standard face image is higher than the preset matching degree threshold, for example, when the matching degree is higher than 80%, the image processing device obtains the verification result that the face is verified, otherwise The image processing device obtains the face based on the verification result to unlock the face, thereby completing the verification result that the image processing verification fails.

S1032. Realize the unlocking of the device based on the verification result, thereby completing image processing.

In S1032, the image processing apparatus may determine whether the device can be unlocked based on the obtained verification result of the face verification, so as to complete the image processing.

In some embodiments, the face unlocking process may be as shown in FIG. 15. In the face input process of FIG. 15, the terminal may perform image quality control on the captured image to be recorded that contains the owner's face, so as to avoid over-captured images. An image that is dark, too bright, or contains an unclear and incomplete face; then in the face detection process, the image processing device uses the method in the embodiment of the application to extract the host’s face from the image to be entered as a standard Face image, and then through the face alignment process, automatically locate the key facial feature points from the standard facial image, such as eyes, nose tip, mouth corner points, eyebrows, and contour points of various parts of the face, and perform live detection of the key facial feature points. Make sure to enter a real person to prevent mask attacks; finally, extract features from the key facial feature points that have passed live detection, and pre-store the extracted features as standard facial features on the terminal or server. For the face unlocking process, the terminal may perform the above image quality control process in the collected images to be unlocked, and then use the method in the embodiment of this application to extract from the image to be unlocked during the face detection process through the image processing device Extract the target face and align the target face; and perform the line of sight/gaze detection process according to the key feature points of the target face obtained by the face alignment to ensure that the target person is currently looking at the screen; The same vitality detection and feature extraction process is carried out through the key features of the target face detected by the gaze/gaze detection, and finally the corresponding target facial features of the target face are obtained. Finally, the image processing device can perform face comparison based on the target facial features and standard facial features to determine whether the target face is the owner himself, if it is the owner, it can unlock according to the target face, if not the owner You can refuse to use the target face to unlock the face, indicating that the unlocking fails.

Hereinafter, an exemplary application of the face detection process in the face unlocking scene of the embodiment of the present application will be described with reference to FIG. 16.

S001: Perform face detection on the collected 640*400 image to obtain at least one face bounding box.

In S001, the image to be processed is a 640*400 image, the at least one detection object is at least one predicted face detected from the 640*400 image, and the at least one object area information is the confidence of the rectangular area occupied by the at least one predicted face The at least one face bounding box is a rectangular bounding box corresponding to the at least one graphically predicted face obtained according to the at least one object area information.

The process of S001 is consistent with the description of S101, and will not be repeated here.

S002: Eliminate face bounding boxes with a confidence level of less than 0.4, and obtain N remaining bounding boxes.

In S002, 0.4 is a preset reliability threshold, and the N remaining bounding boxes are N remaining object regions. The process of S002 is consistent with the description of S201, and will not be repeated here.

S003. When N is greater than 2, reserve the first bounding box with the largest area and the second bounding box with the second largest area in the N remaining bounding boxes.

In S003, the first bounding box is the first object area, and the second bounding box is the second object area. The process of S003 is consistent with the description of S2021, and will not be repeated here.

S004: Determine whether the area of the second bounding box is less than 640, if yes, execute S005, otherwise, execute S006.

In S004, 640 is a preset area threshold.

S005: Exclude the second bounding box, and use the first bounding box as the target bounding box.

In S005, the target bounding box is the target object area, and the process of S005 is consistent with the description of S301, and will not be repeated here.

S006: Determine whether the area ratio of the second bounding box to the first bounding box is less than 0.36, if yes, execute S007; otherwise, execute S008.

In S006, 0.36 is the preset proportion threshold.

S007: Exclude the second bounding box, and use the first bounding box as the target bounding box.

The process of S007 is consistent with the description of S304, and will not be repeated here.

S008. Calculate L1 and L2, where L1=|x1-320| and L2=|x2-320|.

In S008, x1 is the abscissa of the center point of the first bounding box, x2 is the abscissa of the center point of the second bounding box, L1 is the first distance, and L2 is the second distance. The process of S008 is consistent with the description of S3031. Go into details again.

S009: Determine whether any of L1 is greater than 1.5*w1 and L2 is greater than 1.5*w2 is true, if yes, execute S010, otherwise, execute S011.

In S009, w1 is the or width of the first bounding box, w2 is the width of the second bounding box, 1.5*w1 is the preset distance threshold corresponding to the first bounding box, and 1.5*w2 is the preset distance corresponding to the second bounding box Threshold.

S010: When L1 is greater than 1.5*w1, exclude the first bounding box and use the second bounding box as the target bounding box; when L2 is greater than 1.5*w2, exclude the second bounding box and use the first bounding box as the target bounding box.

The process of S010 is consistent with the description of S3032, and will not be repeated here.

S011. When L1 is less than 1.5*w1 and L2 is less than 1.5*w2, if L1 is greater than L2, the first bounding box is excluded and the second bounding box is taken as the target bounding box; if L2 is greater than L1, the second bounding box is excluded, and the first A bounding box serves as the target bounding box.

The process of S011 is consistent with the description of S3033, and will not be repeated here.

It is understandable that, in the embodiments of the present application, in the embodiments of the present application, the image processing device may successively filter at least one face bounding box through the three dimensions of confidence, size, and position, and finally retain the confidence. The face bounding box that is tall, large in area, and closer to the center of the image to be processed is used as the target bounding box, thereby improving the accuracy of locating the target bounding box, thereby improving face recognition and face unlocking based on the target bounding box And so on the accuracy of image processing.

The following will continue to describe the exemplary structure of the image processing device 455 implemented as a software module provided by the embodiments of the present application. In some embodiments, as shown in FIG. 3, the software module stored in the image processing device 455 of the memory 450 may include :

The target detection module 4551 is configured to perform target detection on the image to be processed according to a preset object category to obtain at least one object area information corresponding to at least one detection object;

The screening module 4552 is configured to screen the at least one detection object according to the at least one object area information, and determine the target detection object from the at least one detection object; the object area information of the target detection object conforms to image processing The required preset standard area information;

The image processing module 4553 is used to perform image processing on the target detection object.

In some embodiments, each object area information in the at least one object area information includes at least one of size, position, and confidence; the filtering module 4552 is further configured to determine from at least one of the at least one object area information Exclude object areas whose confidence, size, or position does not meet the preset standard area information in one object area, thereby determining the target object area; the at least one object area is an area range characterized by the at least one object area information; The detection object in the target object area is used as the target detection object.

In some embodiments, the preset standard area information includes a preset confidence threshold, and the screening module 4552 is further configured to exclude the confidence level from being less than the preset confidence level in the at least one object area Threshold object areas, thereby obtaining N remaining object areas; when N is greater than the preset number threshold, exclude object areas whose size or position does not meet the preset standard area information from the N remaining object areas, thereby determining Out of the target object area; the preset number threshold is a non-zero positive integer.

In some embodiments, the screening module 4552 is further configured to calculate the area of each remaining object area in the N remaining object areas according to the size of the N remaining object areas, so as to determine the N remaining object areas. A first object area with a first large area and a second object area with a second large area in the remaining objects; object areas with an area smaller than the second object area are excluded from the N remaining object areas; from the first object area The object area and the second object area are excluded from the object area whose area or position does not meet the preset standard area information, so as to determine the target object area.

In some embodiments, the preset standard area information includes a preset area threshold and a preset proportion threshold. The filtering module 4552 is further configured to: when the area of the second object area is less than the preset area threshold When the first object area is determined as the target object area; when the area of the first object area and the area of the second object area are both greater than the preset area threshold, the second object area is determined Whether the area ratio of the object area to the first object area is greater than a preset proportion threshold; when the area ratio is greater than the preset proportion threshold, the first object area and the second object area The object area whose position does not meet the preset standard area information is excluded from the data, so as to determine the target object area.

In some embodiments, the screening module 4552 is further configured to determine whether the area ratio of the second object area to the first object area is greater than a preset proportion threshold, and when the area ratio is less than When the proportion threshold is preset, the first object area is determined as the target object area.

In some embodiments, the preset standard area information includes a preset distance threshold, and the filtering module 4552 is further configured to determine the location of the first object area and the location of the second object area respectively according to the location where the first object area is located and the location where the second object area is located. Calculate the first distance from the first object area to the image center of the image to be processed and the second distance from the second object area to the image center; when the first distance is in the second distance When any one of is greater than the preset distance threshold, an object area that is less than the preset distance threshold from the center of the image is taken as the target object area; when the first distance and the second distance are both less than the preset distance threshold When the distance threshold is set, it is determined that the object area corresponding to the smallest distance in the first distance and the second distance is the target object area.

In some embodiments, the screening module 4552 further includes a calculation sub-module configured to determine the first abscissa corresponding to the first object area through the location of the first object area; The position where the first object area is located is the coordinate of the center point of the first object area; the second abscissa corresponding to the second object area is determined by the position of the second object area; and the first object area is calculated. A first lateral distance from an abscissa to the abscissa of the image center point of the image to be processed, and the ratio of the first lateral distance to the width of the first object area is used as the first distance; A second lateral distance from the second abscissa to the abscissa of the image center point is calculated, and the ratio of the second lateral distance to the width of the second object area is used as the second distance.

In some embodiments, the image processing device 455 further includes a prompting module configured to calculate each remaining object area in the N remaining object areas according to the size of the N remaining object areas. After the area of the object area, when the area of each of the remaining object areas is less than the preset area threshold, the image processing flow is ended, prompting that no valid target is detected.

In some embodiments, the prompt module is further configured to calculate the images from the first object area to the to-be-processed image according to the positions of the first object area and the second object area. After the first distance from the center and the second distance from the second object area to the center of the image, when the first distance and the second distance are both greater than the preset distance threshold, the image processing flow is ended, prompting no A valid target was detected.

In some embodiments, when the preset object category is a human face category, the target detection object is a target human face, and the image processing module 4553 is further configured to detect the target face according to a pre-entered standard human face. Perform face verification on the face to obtain a verification result; unlock the face based on the verification result, thereby completing image processing.

It should be noted that the description of the above device embodiment is similar to the description of the above method embodiment, and has similar beneficial effects as the method embodiment. For technical details not disclosed in the device embodiment of the present invention, please refer to the description of the method embodiment of the present invention for understanding.

The embodiments of the present application provide a computer program product or computer program. The computer program product or computer program includes computer instructions, and the computer instructions are stored in a computer-readable storage medium. The processor of the computer device reads the computer instruction from the computer-readable storage medium, and the processor executes the computer instruction, so that the computer device executes the image processing method described in the embodiment of the present application.

The embodiment of the present application provides a computer-readable storage medium storing executable instructions, and the executable instructions are stored therein. When the executable instructions are executed by a processor, the processor will cause the processor to execute the method provided in the embodiments of the present application, for example, , As shown in Figures 4, 7, 10, 11, 12, 13, 14, and 16.

In some embodiments, the computer-readable storage medium may be FRAM, ROM, PROM, EPROM, EEPROM, flash memory, magnetic surface memory, optical disk, or CD-ROM; it may also include one or any combination of the foregoing memories. Various equipment.

In some embodiments, the executable instructions may be in the form of programs, software, software modules, scripts or codes, written in any form of programming language (including compiled or interpreted languages, or declarative or procedural languages), and their It can be deployed in any form, including being deployed as an independent program or as a module, component, subroutine or other unit suitable for use in a computing environment.

As an example, executable instructions may but do not necessarily correspond to files in the file system, and may be stored as part of files that store other programs or data, for example, in a HyperText Markup Language (HTML, HyperText Markup Language) document One or more scripts in are stored in a single file dedicated to the program in question, or in multiple coordinated files (for example, a file storing one or more modules, subroutines, or code parts).

As an example, executable instructions can be deployed to be executed on one computing device, or on multiple computing devices located in one location, or on multiple computing devices that are distributed in multiple locations and interconnected by a communication network Executed on.

In summary, through the embodiments of the present application, the image processing device can filter at least one object region information layer by layer through the three dimensions of confidence, size, and position, and finally retain high confidence, large area, and high reliability. The object area closer to the center of the image to be processed is used as the target object area, thereby improving the accuracy of locating the target object area, and further improving the accuracy of image processing based on the target object area.

The above are only examples of the present application, and are not used to limit the protection scope of the present application. Any modification, equivalent replacement and improvement made within the spirit and scope of this application are all included in the protection scope of this application.

Industrial applicability

In the embodiment of the present application, the image processing device may successively filter the information of at least one object area through the three dimensions of confidence, size, and position, and finally retain high confidence, large area, and more distance from the center of the image to be processed. The near object area is used as the target object area, thereby improving the accuracy of locating the target object area, and further improving the accuracy of image processing based on the target object area.

Claims

An image processing method, including:

Performing target detection on the image to be processed according to the preset object category, to obtain at least one object region information corresponding to the at least one detection object;

According to the at least one object area information, the at least one detection object is screened, and the target detection object is determined from the at least one detection object; the object area information of the target detection object meets the preset standard required by image processing Area information

Image processing is performed on the target detection object.
The method according to claim 1, wherein each object area information in the at least one object area information includes at least one of size, position, and confidence;

The screening the at least one detection object according to the at least one object region information, and determining the target detection object from the at least one detection object includes:

According to the at least one object area information, exclude from the at least one object area an object area whose confidence, size, or position does not meet the preset standard area information, so as to determine the target object area; the at least one object area is all The area range represented by the at least one object area information;

The detection object in the target object area is used as the target detection object.
2. The method according to claim 2, wherein the preset standard area information includes a preset confidence threshold, and the at least one object area is excluded from the at least one object area in terms of confidence, size, or location according to the at least one object area information. The object area that meets the preset standard area information to determine the target object area includes:

In the at least one object area, exclude object areas whose confidence is less than the preset confidence threshold, so as to obtain N remaining object areas;

When N is greater than the preset number threshold, exclude the object areas whose size or position does not meet the preset standard area information from the N remaining object areas, so as to determine the target object area; the preset number threshold Is a non-zero positive integer.
The method according to claim 3, wherein the excluding from the N remaining object areas an object area whose size or position does not meet the preset standard area information, so as to determine the target object area, comprises:

According to the size of the N remaining object areas, the area of each remaining object area in the N remaining object areas is calculated, so as to determine the first object area and the second object area with the first large area among the N remaining objects. Large area of the second object area;

Excluding from the N remaining object areas an object area with an area smaller than the second object area;

Excluding from the first object area and the second object area an object area whose area or position does not meet the preset standard area information, thereby determining the target object area.
The method according to claim 4, wherein the preset standard area information includes a preset area threshold value and a preset proportion threshold value, and the excluding area from the first object area and the second object area or The object area whose position does not conform to the preset standard area information to determine the target object area includes:

When the area of the second object area is smaller than the preset area threshold, determining the first object area as the target object area;

When the area of the first object area and the area of the second object area are both greater than the preset area threshold, it is determined whether the area ratio of the second object area to the first object area is greater than a preset value Percentage threshold;

When the area ratio is greater than the preset proportion threshold, the object area whose position does not meet the preset standard area information is excluded from the first object area and the second object area, thereby determining the Target area.
5. The method according to claim 5, wherein after determining whether the area ratio of the second object area to the first object area is greater than a preset proportion threshold, the method further comprises:

When the area ratio is less than a preset proportion threshold, the first object area is determined as the target object area.
The method according to claim 5, wherein the preset standard area information includes a preset distance threshold, and the excluded position from the first object area and the second object area does not meet the preset standard The object area of the area information to determine the target object area includes:

According to the location of the first object area and the location of the second object area, the first distance from the first object area to the image center of the image to be processed and the distance from the second object area to the The second distance from the center of the image;

When any one of the first distance and the second distance is greater than the preset distance threshold, use an object area that is less than the preset distance threshold from the center of the image as the target object area;

When the first distance and the second distance are both less than a preset distance threshold, it is determined that the object area corresponding to the smallest distance in the first distance and the second distance is the target object area.
8. The method according to claim 7, wherein the image from the first object area to the to-be-processed image is respectively calculated according to the location of the first object area and the location of the second object area The first distance from the center and the second distance from the second object area to the image center include:

Determine the first abscissa corresponding to the first object area by the location of the first object area; the location of the first object area is the coordinate of the center point of the first object area;

Determine the second abscissa corresponding to the second object area by using the position where the second object area is located;

Calculate the first lateral distance between the first abscissa and the abscissa of the image center point of the image to be processed, and use the ratio of the first lateral distance to the width of the first object area as the First distance

A second lateral distance from the second abscissa to the abscissa of the image center point is calculated, and the ratio of the second lateral distance to the width of the second object area is used as the second distance.
The method according to claim 4, wherein after the calculating the area of each of the N remaining object areas according to the size of the N remaining object areas, the method further comprises:

When the area of each remaining object area is less than the preset area threshold, the image processing flow is ended, and it is prompted that no valid target is detected.
8. The method according to claim 7, wherein the first object area from the first object area to the image center of the image to be processed is calculated according to the positions of the first object area and the second object area. After the distance and the second distance from the second object area to the center of the image, the method further includes:

When the first distance and the second distance are both greater than the preset distance threshold, the image processing flow is ended, and a prompt is not detected for a valid target.
The method according to any one of claims 1-8, wherein when the preset object category is a face category, the target detection object is a target human face, and the image processing is performed on the target detection object, include:

Perform face verification on the target face according to the pre-entered standard face, and obtain a verification result;

The face is unlocked based on the verification result, thereby completing image processing.
An image processing device, including:

The target detection module is configured to perform target detection on the image to be processed according to a preset object category to obtain at least one object area information corresponding to at least one detection object;

The screening module is configured to screen the at least one detection object according to the at least one object area information, and determine the target detection object from the at least one detection object; the object area information of the target detection object conforms to the image processing institute The required preset standard area information;

The determining module is used to perform image processing on the target detection object.
An image processing device, including:

Memory, used to store executable instructions;

The processor is configured to implement the method according to any one of claims 1 to 11 when executing the executable instructions stored in the memory.
A computer-readable storage medium storing executable instructions for implementing the method according to any one of claims 1 to 11 when executed by a processor.