Search

CN-122022737-A - Unmanned hotel management method and system

CN122022737ACN 122022737 ACN122022737 ACN 122022737ACN-122022737-A

Abstract

The invention relates to the technical field of face recognition and discloses an unmanned hotel management method and system, wherein the unmanned hotel management method comprises the following steps of S101, S102, S103, S104, S105, S106, and S106, wherein a standardized frame is obtained through calculation, a local observation area is determined, S103, an external nasal valve boundary curve is determined, S104, the geometric curvature of the external nasal valve boundary curve is calculated, S105, the second-order energy density of the collapse curvature of the external nasal valve is calculated, and an admission token is generated. The invention eliminates geometric distortion caused by small head movement by selecting the human face datum point to perform similar transformation alignment, contours a local observation area according to the human face anatomical proportion, eliminates irrelevant interference, extracts a boundary curve through gradient energy line integration, combines energy aggregation and normalization of second derivative of curvature time, avoids the risk of prosthesis attack, and improves the safety and stability of human face recognition under an unmanned self-help scene.

Inventors

  • MA XIANJUN

Assignees

  • 安徽云视通讯有限公司

Dates

Publication Date
20260512
Application Date
20260212

Claims (10)

  1. 1. An unmanned hotel management method, comprising the steps of: Step S101, collecting a video sequence prompting a user to inhale, detecting nose tips, human middle points and inter-eyebrow points as human face stability reference points, and uniformly aligning the video sequence to a first frame coordinate system by utilizing least square similarity transformation to obtain a standardized frame; Step S102, a local observation area containing an external nasal valve structure is defined in a standardized frame according to the anatomical proportion of the face, so that the anatomical position of an observation object is ensured to be constant; Step S103, calculating an image gradient module in a local observation area, and searching a path for maximizing a gradient energy line integral to determine an external nasal valve boundary curve; Step S104, calculating the geometric curvature of the boundary curve of the external nasal valve, and calculating the time second derivative of the curvature by utilizing the center difference operation so as to quantify the acceleration level change of the boundary shape; Step S105, carrying out energy aggregation and normalization on the time second derivative of curvature in a space-time domain, calculating the outer nasal valve collapse curvature second-order energy density, and comparing the outer nasal valve collapse curvature second-order energy density with a preset threshold value to output a living body judgment result; and S106, selecting a maximum frame of the gradient energy line integral to extract a face vector only when the living body judging result is passing, binding the face vector with the guest room entrance record and generating an admission token.
  2. 2. The method for managing the unmanned hotel according to claim 1, wherein a time sequence comprising a plurality of sampling moments is established, adjacent frame time intervals are set to a fixed value, the total acquisition duration is set to an acquisition window length, and an original frame set formed by an image intensity function is acquired; Positioning the nose tip, the human midpoint and the inter-eyebrow point in the original frame image corresponding to each sampling moment, and defining the nose tip, the human midpoint and the inter-eyebrow point as two-dimensional plane coordinate vectors; Constructing a similar transformation model comprising scale parameters, a two-dimensional rotation matrix and translation vectors; Solving optimal similarity transformation parameters by taking the square sum of Euclidean distances between the nose tip, the middle point of a person and the reference points corresponding to the eyebrow points and the initial sampling moment after the transformation at the current sampling moment as a target, wherein the two-dimensional rotation matrix needs to meet the constraint that the orthogonality constraint and the determinant are one; And carrying out coordinate inverse transformation and resampling on the original frame image at the current sampling moment by using inverse mapping of the similar transformation model, and generating a standardized frame sequence aligned with the initial sampling moment coordinate system.
  3. 3. The method of claim 1, wherein receiving the standardized frame sequence and the nose tip of the first frame, and setting a horizontal unit vector in a first frame coordinate system; setting an inner radius, an outer radius, an inner angle and an outer angle according to the anatomical proportion of the face, wherein the inner radius is smaller than the outer radius, and the inner angle is smaller than the outer angle and smaller than ninety degrees; Defining a local observation area in a first frame coordinate system, and defining that pixel points in the local observation area meet distance constraint and angle constraint; the distance constraint prescribes that the Euclidean distance between the pixel point and the nose tip of the first frame is between the inner radius and the outer radius, and the angle constraint prescribes that the included angle between the connecting line vector of the pixel point relative to the nose tip of the first frame and the horizontal unit vector is in a positive angle interval or a negative angle interval determined by the inner angle and the outer angle; And outputting the local observation area for calculating the boundary curve of the external nasal valve.
  4. 4. The method of unmanned hotel management of claim 1, wherein a standardized sequence of frames and local viewing areas are received; calculating a lateral coordinate partial derivative and a longitudinal coordinate partial derivative for each frame of standardized image in the local observation area, and calculating the arithmetic square root of the sum of the square of the lateral coordinate partial derivative and the square of the longitudinal coordinate partial derivative to obtain an image gradient module; Defining a candidate curve family positioned in a local observation area, wherein curves in the candidate curve family meet continuous micro-condition and unit tangent vector modular length constraint; constructing a target functional, and defining the target functional as the path integral of the image gradient module along the candidate curve; Traversing candidate curve family to search the curve maximizing the objective function value, determining the curve maximizing the objective function value as the outer nasal valve boundary curve, and outputting the outer nasal valve boundary curve and the image gradient mode.
  5. 5. The unmanned hotel management method according to claim 1, wherein the external nasal valve boundary curve sequence is received, arc length parameterization description is adopted for the boundary curve, the total arc length and the arc length parameter value range are determined, and time discretization setting is adopted; For each sampling moment, calculating the first derivative and the second derivative of the boundary curve coordinates with respect to arc length parameters, constructing a geometric curvature calculation formula, setting a numerator as the product of the first derivative of the transverse coordinates and the second derivative of the longitudinal coordinates minus the product of the first derivative of the longitudinal coordinates and the second derivative of the transverse coordinates, setting an denominator as the third power of the sum of the square of the first derivative of the transverse coordinates and the square of the first derivative of the longitudinal coordinates, and obtaining the geometric curvature according to the geometric curvature calculation formula.
  6. 6. The method of unmanned hotel management according to claim 5, wherein the time second derivative of the geometric curvature is calculated using a central difference operation, the sum of the geometric curvature at the next sampling time and the geometric curvature at the previous sampling time is calculated, the twice of the geometric curvature at the current sampling time is subtracted, the result is divided by the square of the time interval of the adjacent frames, the calculation result is defined as the time second derivative of the curvature, and the time-space sequence of the geometric curvature and the time second derivative of the curvature is output.
  7. 7. The method of claim 1, wherein receiving a sequence of time second derivatives of curvature, determining total arc length of outer nasal valve boundary, determining adjacent frame time intervals; The effective time window length is defined as the product of the number of time steps that can be center differentiated and the time interval of adjacent frames, and the square value of the time second derivative of the curvature is calculated as the energy term.
  8. 8. The method for managing the unmanned hotel according to claim 7, wherein the integration is performed firstly for the total arc length of the energy item along the boundary of the external nasal valve, and then the discrete integration is performed for the integration result in an effective time window to obtain the total energy value; the total energy value is divided by the product of the total arc length of the boundary of the external nasal valve and the length of the effective time window, and the space and time normalization processing is completed, so that the second-order energy density of the collapse curvature of the external nasal valve is obtained; The method comprises the steps of comparing the collapse curvature second-order energy density of the outer nasal valve with a preset threshold, outputting a passing judgment result if the collapse curvature second-order energy density of the outer nasal valve is larger than or equal to the preset threshold, and outputting a refusal judgment result if the collapse curvature second-order energy density of the outer nasal valve is smaller than the preset threshold.
  9. 9. The unmanned hotel management method according to claim 1, wherein the living body determination result is received, and if the living body determination result is passed, the steps of; For each frame in the standardized frame sequence, calculating the path integral of the image gradient mode along the outer nasal valve boundary curve by utilizing the arc length parameterization form of the outer nasal valve boundary curve, and defining the path integral as gradient energy line integral; traversing the gradient energy line integrals corresponding to all frames in the acquisition window, selecting the frame with the maximum gradient energy line integral value, and determining the frame with the maximum gradient energy line integral value as the clearest frame; extracting features of the clearest frame by using a feature mapping function to generate a real number domain multi-dimensional face vector; The method comprises the steps of constructing a registration record, wherein the registration record comprises a resident identification, a face vector, a room number and a valid period, encoding the registration record by using an encoding mapping function to generate a permission token, and outputting the face vector, the registration record and the permission token.
  10. 10. An unmanned hotel management system, wherein an unmanned hotel management method as claimed in any of claims 1 to 9 is performed, comprising: The standardized frame calculation module is used for collecting video sequences prompting a user to inhale, detecting nose tips, human middle points and eyebrow points as human face stability reference points, and uniformly aligning the video sequences to a first frame coordinate system by utilizing least square similarity transformation to obtain a standardized frame; the local observation area demarcation module is used for demarcating a local observation area containing an external nasal valve structure in a standardized frame according to the face anatomical proportion, so as to ensure the anatomical position of an observation object to be constant; the outer nasal valve boundary curve determining module calculates an image gradient module in the local observation area, and searches a path for maximizing gradient energy line integral so as to determine an outer nasal valve boundary curve; The geometric curvature calculation module calculates the geometric curvature of the boundary curve of the external nasal valve, and calculates the time second derivative of the curvature by utilizing the center difference operation so as to quantify the acceleration level change of the boundary shape; The second-order energy density calculation module is used for carrying out energy aggregation and normalization on the time second derivative of the curvature in a space-time domain, calculating the outer nasal valve collapse curvature second-order energy density, and comparing the outer nasal valve collapse curvature second-order energy density with a preset threshold value to output a living body judgment result; And the admission token generation module is used for selecting the maximum frame of the gradient energy line integral to extract the face vector only when the living body judging result is passing, binding the face vector with the guest room admission record and generating the admission token.

Description

Unmanned hotel management method and system Technical Field The invention relates to the technical field of face recognition, in particular to an unmanned hotel management method and system. Background With the intelligent development of hotels, unmanned self-help check-in modes are gradually popularized. The face recognition technology is used as a core identity verification means of the mode, and the safety and accuracy of the face recognition technology directly influence the operation effect of the system. In the prior art, static face features are mostly adopted for identity comparison, and living body judgment is implemented without combining with physiological motion features of the face. The method only extracts the plane texture or contour information of the human face, and cannot distinguish the false body such as a real human body from a photo video. Lawless persons can break through verification by forging face prostheses, so that potential safety hazards of guest rooms are caused. Meanwhile, in the face image acquisition process, small-amplitude translational rotation or scaling of the head of the user is easy to occur. These movements result in significant geometric distortion between acquisition frames. The stability of the reference point selected by the traditional image alignment method is insufficient, and the distortion influence cannot be effectively eliminated. The interframe geometric differences can reduce the accuracy of the subsequent steps of outer nasal valve boundary extraction curvature calculation, etc. The physiological motion characteristics depending on the living body judgment cannot be accurately quantified, and the reliability of the judgment result is further affected. In addition, the existing method does not effectively combine the physiological motion characteristics and the identity characteristics of the living body. Under an unattended scene, the verification safety cannot be guaranteed, and the accuracy of identity recognition is difficult to improve. The problems limit popularization and application of the hotel unmanned self-help check-in face recognition system together. Disclosure of Invention The invention provides an unmanned hotel management method and system, which solve the technical problems in the background technology. The invention provides an unmanned hotel management method, which comprises the following steps: Step S101, collecting a video sequence prompting a user to inhale, detecting nose tips, human middle points and inter-eyebrow points as human face stability reference points, and uniformly aligning the video sequence to a first frame coordinate system by utilizing least square similarity transformation to obtain a standardized frame; Step S102, a local observation area containing an external nasal valve structure is defined in a standardized frame according to the anatomical proportion of the face, so that the anatomical position of an observation object is ensured to be constant; Step S103, calculating an image gradient module in a local observation area, and searching a path for maximizing a gradient energy line integral to determine an external nasal valve boundary curve; Step S104, calculating the geometric curvature of the boundary curve of the external nasal valve, and calculating the time second derivative of the curvature by utilizing the center difference operation so as to quantify the acceleration level change of the boundary shape; Step S105, carrying out energy aggregation and normalization on the time second derivative of curvature in a space-time domain, calculating the outer nasal valve collapse curvature second-order energy density, and comparing the outer nasal valve collapse curvature second-order energy density with a preset threshold value to output a living body judgment result; and S106, selecting a maximum frame of the gradient energy line integral to extract a face vector only when the living body judging result is passing, binding the face vector with the guest room entrance record and generating an admission token. The invention provides an unmanned hotel management system, which comprises: The standardized frame calculation module is used for collecting video sequences prompting a user to inhale, detecting nose tips, human middle points and eyebrow points as human face stability reference points, and uniformly aligning the video sequences to a first frame coordinate system by utilizing least square similarity transformation to obtain a standardized frame; the local observation area demarcation module is used for demarcating a local observation area containing an external nasal valve structure in a standardized frame according to the face anatomical proportion, so as to ensure the anatomical position of an observation object to be constant; the outer nasal valve boundary curve determining module calculates an image gradient module in the local observation area, and searches a path for maximizing gradient energy lin