Face recognition method and system for movable sentry robot

Abstract

The invention discloses a face recognition method and system for a movable sentry robot, and the method comprises the steps 1-4: 1, obtaining a face image detected by the movable sentry robot; 2, extracting contour information in the face image, and obtaining a face contour of an unshielded part according to standard features of the face; 3, extracting key points of the facial contour of the unshielded part, and segmenting the facial contour of the unshielded part into line segments according to the key points; and 4, extracting functions corresponding to the line segments to obtain a function set, searching personnel information corresponding to the function set in a database, and outputting the personnel information. According to the method, the contour is converted into the line segments, so that the shielded face is recognized, the accuracy of face recognition is improved, the accuracy of face requirements of face recognition is also improved, and the user threshold of face recognition is reduced.

Description

Technical Field

[0001] The present invention relates to the field of information technology, and in particular to a face recognition method and system for a mobile sentinel robot. Background Art

[0002] On construction sites, construction is currently managed intelligently, especially when inspecting personnel on the construction site. Facial recognition is generally used, which can quickly scan the characteristics of the personnel.

[0003] The current face recognition technology generally scans the facial features (eyes, nose, mouth, etc.), facial contours, pupil position, size and brightness, etc., and then determines the facial information through the appropriate relationship among them to complete the identification of the person. Although this method can quickly identify the person, it has an obvious disadvantage, that is, when performing face recognition, there cannot be any obstruction of the face. Therefore, when performing face recognition, due to the presence of obstructions, the information for face recognition will be lost, resulting in the inability to perform face recognition. Summary of the invention

[0004] The technical problem to be solved by the present invention is to provide a face recognition method and system for a mobile sentinel robot in view of the deficiencies in the above-mentioned prior art, so as to realize the recognition of the obstructed face by converting the contour into line segments.

[0005] In order to solve the above technical problems, the technical solution adopted by the present invention is: a face recognition method and system for a mobile sentinel robot.

[0006] A face recognition method for a mobile sentinel robot comprises the following steps:

[0007] Step 1: Obtain the face image detected by the mobile sentinel robot;

[0008] Step 2: extracting contour information from the face image, and obtaining the facial contour of the unobstructed part according to standard features of the face;

[0009] Step 3: extracting key points of the facial contour of the unobstructed part, and segmenting the facial contour of the unobstructed part into individual line segments according to the key points;

[0010] Step 4: extracting the functions corresponding to the line segments to obtain a function set, searching the database for personnel information corresponding to the function set, and outputting the personnel information;

[0011] The database is used to store personnel information and a function set corresponding to the personnel information.

[0012] Further, in step 2, obtaining the facial contour of the unobstructed part according to the standard features of the face includes the following steps:

[0013] Extracting facial features contained in the contour information, and comparing the facial features contained in the contour information with standard features to determine the types of the facial features contained in the contour information;

[0014] Obtaining an unobstructed portion of the contour information according to the types of the included facial features;

[0015] The facial contour of the unobstructed portion in the face image is obtained according to the contour position corresponding to the unobstructed portion in the contour information.

[0016] Furthermore, the standard features include the positions of facial features of the person.

[0017] Furthermore, in step 2, extracting the contour information in the face image includes the following steps:

[0018] Obtaining the pixel value of each pixel point of the face image;

[0019] When the difference between the pixel values ​​of two adjacent pixel points is greater than a first set value, the pixel point is marked;

[0020] Traversing each pixel in the face image to obtain all marked pixels;

[0021] All pixel points are processed and output to obtain the contour information in the face image.

[0022] Furthermore, the processing of all pixels to obtain contour information in the face image includes the following steps:

[0023] Obtaining the positions of all the marked pixel points to obtain a preliminary outline;

[0024] Remove isolated pixels from the preliminary contour to obtain an advanced contour;

[0025] According to the contour positions corresponding to the types of various facial features in the standard features, the pixel points outside the contour positions corresponding to the types of various facial features in the advanced contour are removed to obtain the contour information in the facial image.

[0026] Further, in step 3, extracting key points of the facial contour of the unobstructed part includes:

[0027] The points in the facial contour of the unobstructed part whose angles are greater than the set angle are taken as the key points.

[0028] Furthermore, taking the point in the facial contour of the unobstructed part with an angle greater than a set angle as the key point comprises the following steps:

[0029] Extracting the coordinates of each pixel point in the facial contour of the unobstructed part;

[0030] Calculating the slope between the coordinates of two adjacent pixel points of the facial contour of the unblocked part;

[0031] When the difference between two adjacent slopes is greater than a second set value, the middle pixel point among the three pixel points is identified as the key point;

[0032] Output the key points.

[0033] A face recognition system for a mobile sentinel robot, having the same inventive concept as the above method, comprises:

[0034] An image acquisition module, used to acquire a face image detected by the mobile sentinel robot;

[0035] A contour extraction module, used to extract contour information in the face image and obtain the facial contour of the unobstructed part according to standard features of the face;

[0036] A line segment extraction module, used to extract key points of the facial contour of the unobstructed part, and segment the facial contour of the unobstructed part into individual line segments according to the key points;

[0037] A personnel confirmation module is used to extract the functions corresponding to each of the line segments to obtain a function set, search the database for personnel information corresponding to the function set, and output the personnel information;

[0038] The database is used to store personnel information and a function set corresponding to the personnel information.

[0039] Compared with the prior art, the present invention has the following advantages:

[0040] The present invention provides a face recognition method and system for a mobile sentinel robot, which collects and processes face images, extracts facial contours of people in the image, converts the facial contours into line segments, and then represents the line segments in a function manner, thereby accurately determining the person information according to the person and the corresponding function. Compared with the traditional method of identifying and locating people by facial features, the method is more accurate. At the same time, when identifying people, even if there is some area occlusion on the face of the person, the person can be identified by the function of part of the contour, which further improves the accuracy of face recognition, improves the accuracy required for face recognition, and reduces the user threshold of face recognition.

[0041] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 A schematic diagram of the overall process of a face recognition method for a mobile sentinel robot provided by the present invention.

[0043] Figure 2 A schematic diagram of the overall structure of a face recognition system for a mobile sentinel robot provided by the present invention. DETAILED DESCRIPTION

[0044] like Figure 1-2 As shown, the present invention provides a face recognition method and system for a mobile sentinel robot, wherein a face recognition system for a mobile sentinel robot is a supporting hardware practice method for the face recognition method of a mobile sentinel robot, and both are based on the same inventive concept.

[0045] like Figure 1 As shown, the present invention proposes a face recognition method for a mobile sentinel robot, comprising the following steps:

[0046] Step 1: Obtain the face image detected by the mobile sentinel robot;

[0047] Step 2: extracting contour information from the face image, and obtaining the facial contour of the unobstructed part according to standard features of the face;

[0048] Step 3: extracting key points of the facial contour of the unobstructed part, and segmenting the facial contour of the unobstructed part into individual line segments according to the key points;

[0049] Step 4: extracting the functions corresponding to the line segments to obtain a function set, searching the database for personnel information corresponding to the function set, and outputting the personnel information;

[0050] The database is used to store personnel information and a function set corresponding to the personnel information.

[0051] In a construction site, a mobile sentinel robot performs mobile management within the construction site. Generally, the mobile sentinel is equipped with a camera, through which facial images can be obtained, and then the method provided by the present invention is used to identify personnel information and determine whether the person is a person on the construction site.

[0052] When the mobile sentinel robot moves on the construction site, it will automatically extract the face image. In step 1 of the present invention, even if the face image detected by the mobile sentinel robot is obtained, step 2 is to obtain the unobstructed facial contour. For example, if a person wears a mask, the present invention obtains the contour of the part that is not obstructed by the mask, or the face of the person is partially obstructed by leaves, and the present invention recognizes the contour of the part that is not obstructed. Step 3 is to split the contour into multiple line segments, and step 4 is based on the function corresponding to each line segment. Since the function has accuracy, the function corresponding to each person's facial contour is different. Therefore, the present invention uses the function to count and identify the personnel, so as to accurately judge the information of the personnel. Even if the face of the personnel has some area obstruction, the personnel can be identified by the function of the partial contour, further improving the accuracy of face recognition, and also improving the accuracy of face recognition for the face requirements, and reducing the user threshold of face recognition.

[0053] Specifically, in step 2, obtaining the facial contour of the unobstructed part according to the standard features of the face includes the following steps:

[0054] Extracting facial features contained in the contour information, and comparing the facial features contained in the contour information with standard features to determine the types of the facial features contained in the contour information;

[0055] Obtaining an unobstructed portion of the contour information according to the types of the included facial features;

[0056] The facial contour of the unobstructed portion in the face image is obtained according to the contour position corresponding to the unobstructed portion in the contour information.

[0057] The standard features include the positions of the facial features of the person, including eyes, nose, mouth, face shape, etc.

[0058] When identifying, the present invention determines the blocked part according to the occlusion of the facial features, so that the unblocked facial contour can be effectively identified. For example, when people wear masks, they will block the mouth and nose, and then block the facial contour features, and then replace the remaining facial features to obtain the facial contours corresponding to these facial features as the unblocked contours.

[0059] Meanwhile, in step 2, extracting the contour information in the face image includes the following steps:

[0060] Obtaining the pixel value of each pixel point of the face image;

[0061] When the difference between the pixel values ​​of two adjacent pixel points is greater than a first set value, the pixel point is marked;

[0062] Traversing each pixel in the face image to obtain all marked pixels;

[0063] All pixel points are processed and output to obtain the contour information in the face image.

[0064] The present invention determines the contour by the difference between adjacent pixel values. Generally, the area with a large pixel difference is the boundary between the face and the background. Therefore, the present invention performs segmentation based on the difference in pixel values. After traversing all pixel points, the contour information in the face image can be completely extracted, thereby realizing a solution for extracting contour information based on an image.

[0065] At the same time, the process of processing all the pixels and outputting the contour information in the face image comprises the following steps:

[0066] Obtaining the positions of all the marked pixel points to obtain a preliminary outline;

[0067] Remove isolated pixels from the preliminary contour to obtain an advanced contour;

[0068] According to the contour positions corresponding to the types of various facial features in the standard features, the pixel points outside the contour positions corresponding to the types of various facial features in the advanced contour are removed to obtain the contour information in the facial image.

[0069] When the present invention obtains the contour information in the face image, some errors are inevitable due to the use of pixel difference. Therefore, the present invention subsequently removes the errors of the previous step by removing isolated points and obviously erroneous points, thereby obtaining the contour information in the optimized face image.

[0070] In the present invention, in step 3, extracting key points of the facial contour of the unobstructed part includes:

[0071] The points in the facial contour of the unobstructed part whose angles are greater than the set angle are taken as the key points.

[0072] Specifically, the step of taking a point in the facial contour of the unobstructed portion whose angle is greater than a set angle as the key point comprises the following steps:

[0073] Extracting the coordinates of each pixel point in the facial contour of the unobstructed part;

[0074] Calculating the slope between the coordinates of two adjacent pixel points of the facial contour of the unblocked part;

[0075] When the difference between two adjacent slopes is greater than a second set value, the middle pixel point among the three pixel points is identified as the key point;

[0076] Output the key points.

[0077] When removing key points, the present invention uses the turning parts of adjacent line segments, divides the facial contour at the turning points, and then obtains various line segments.

[0078] In addition, in view of the angle problem that may occur when identifying a person, the present invention can make a range adjustment on the coefficients in the function, so that the person's information can be determined as long as the coefficients are within the set range.

[0079] like Figure 2 As shown, based on the above method, the present invention also proposes a face recognition system for a mobile sentinel robot, which is characterized by comprising:

[0080] An image acquisition module, used to acquire a face image detected by the mobile sentinel robot;

[0081] A contour extraction module, used to extract contour information in the face image and obtain the facial contour of the unobstructed part according to standard features of the face;

[0082] A line segment extraction module, used to extract key points of the facial contour of the unobstructed part, and segment the facial contour of the unobstructed part into individual line segments according to the key points;

[0083] A personnel confirmation module is used to extract the functions corresponding to each of the line segments to obtain a function set, search the database for personnel information corresponding to the function set, and output the personnel information;

[0084] The database is used to store personnel information and a function set corresponding to the personnel information.

[0085] Each module executes the above steps 1-4 respectively. Each module is executed in the form of a software module. The execution method includes but is not limited to a processor.

[0086] The above description is only a preferred embodiment of the present invention and does not limit the present invention in any way. Any simple modification, change and equivalent structural change made to the above embodiment based on the technical essence of the present invention still falls within the protection scope of the technical solution of the present invention.

Claims

1. A face recognition method for a mobile sentinel robot, characterized in that: The following steps are involved: Step 1: Obtain the face image detected by the mobile sentinel robot; Step 2: extracting contour information from the face image, and obtaining the facial contour of the unobstructed part according to standard features of the face; Step 3: extracting key points of the facial contour of the unobstructed part, and segmenting the facial contour of the unobstructed part into individual line segments according to the key points; Step 4: extracting the functions corresponding to the line segments to obtain a function set, searching the database for personnel information corresponding to the function set, and outputting the personnel information; The database is used to store personnel information and a function set corresponding to the personnel information.

2. A face recognition method for a mobile sentinel robot according to claim 1, characterized in that: In step 2, obtaining the facial contour of the unobstructed part according to the standard features of the face includes the following steps: Extracting facial features contained in the contour information, and comparing the facial features contained in the contour information with standard features to determine the types of the facial features contained in the contour information; Obtaining an unobstructed portion of the contour information according to the types of the included facial features; The facial contour of the unobstructed portion in the face image is obtained according to the contour position corresponding to the unobstructed portion in the contour information.

3. A face recognition method for a mobile sentinel robot according to claim 2, characterized in that: The standard features include the positions of the facial features of the person.

4. A face recognition method for a mobile sentinel robot according to claim 3, characterized in that: In step 2, extracting the contour information in the face image includes the following steps: Obtaining the pixel value of each pixel point of the face image; When the difference between the pixel values ​​of two adjacent pixel points is greater than a first set value, the pixel point is marked; Traversing each pixel in the face image to obtain all marked pixels; All pixel points are processed and output to obtain the contour information in the face image.

5. A face recognition method for a mobile sentinel robot according to claim 4, characterized in that: The step of processing all the pixels and outputting the contour information in the face image comprises the following steps: Obtaining the positions of all the marked pixel points to obtain a preliminary outline; Remove isolated pixels from the preliminary contour to obtain an advanced contour; According to the contour positions corresponding to the types of various facial features in the standard features, the pixel points outside the contour positions corresponding to the types of various facial features in the advanced contour are removed to obtain the contour information in the facial image.

6. A face recognition method for a mobile sentinel robot according to claim 1, characterized in that: In step 3, extracting key points of the facial contour of the unobstructed part includes: The points in the facial contour of the unobstructed part whose angles are greater than the set angle are taken as the key points.

7. A face recognition method for a mobile sentinel robot according to claim 6, characterized in that: The step of taking a point in the facial contour of the unobstructed portion whose angle is greater than a set angle as the key point comprises the following steps: Extracting the coordinates of each pixel point in the facial contour of the unobstructed part; Calculating the slope between the coordinates of two adjacent pixel points of the facial contour of the unblocked part; When the difference between two adjacent slopes is greater than a second set value, the middle pixel point among the three pixel points is identified as the key point; Output the key points.

8. A face recognition system for a mobile sentinel robot, characterized in that: include: An image acquisition module, used to acquire a face image detected by the mobile sentinel robot; A contour extraction module, used to extract contour information in the face image and obtain the facial contour of the unobstructed part according to standard features of the face; A line segment extraction module, used to extract key points of the facial contour of the unobstructed part, and segment the facial contour of the unobstructed part into individual line segments according to the key points; A personnel confirmation module is used to extract the functions corresponding to each of the line segments to obtain a function set, search the database for personnel information corresponding to the function set, and output the personnel information; The database is used to store personnel information and a function set corresponding to the personnel information.

Images