wherein, S (B)_i,B_j) A similarity measure between the two elements; b is_i，B_jRespectively representing the colour areas represented by the two elements, S_c(B_i,B_j) Representing a color feature metric between two elements; s_r(B_i,B_j) Representing a measure of the luminance characteristic between two elements, S_l(B_i,B_j) Representing the degree of position between two elementsAn amount; and a and b are respectively the weight of the color characteristic and the brightness characteristic, the sum of the weight of the color characteristic and the brightness characteristic is 1, and c is a correction parameter with the value range of 0-10.

If the similarity measurement is smaller than the elements in the preset range, forming a second pixel matrix, and finally combining the second pixel matrix into a region, namely the first segmentation region; then clustering images of the color images by using LAB information through a python algorithm after super-pixel equalization processing, and reducing the gray value of a class in contact with the image boundary in a clustering result to 0; and finally obtaining the first segmentation area.

Because a plurality of contour extraction algorithms exist in the prior art, image information generally needs to be acquired by matching with an illumination light source when light is dim, shadows of people and objects are relatively obvious at the moment, and shadow interference caused by shadow interference or other light reasons can be caused when the gray level images in the first information and the second information are subjected to image extraction; the acquisition method is improved from the viewpoint of vision, the pixel points on the boundary are analyzed, and shadow interference caused by shadows or other light reasons can be effectively eliminated.

S105, matching the personnel information of the first information and the second information within preset time, and judging whether the personnel information is the same target object; if yes, calculating the article information missing in the article information in the second information to obtain first article information, namely the articles stored in the corridor.

The preset time is the time required by the user from entering the corridor to leaving the corridor, generally the preset time is not more than 5min, and in the embodiment, the preset time is set to be 3-5 minutes. The first information and the second information within 3-5 minutes are matched. If the first information and the second information have the same personnel information within the preset time interval; then further analyzing the article information between the first information and the second information, and if the article information between the first information and the second information is the same, indicating that the user does not place any article in the corridor; otherwise, the user is indicated to place the article in the corridor, and then the difference between the article information of the first information and the article information of the second information is further compared and analyzed, and the article information missing in the article information in the second information is calculated, namely the first article information. Taking an electric vehicle as an example, the electric vehicle is a rigid object in the moving process, and the deformation amount is relatively small, so that the first information and the second information can be processed directly by the shooting perspective principle, the first information and the second information are converted to the same visual angle, then the first information and the second information are compared, and a specific comparison method can be referred to in the following of the embodiment.

The matching method of the personnel information comprises the following steps: firstly, acquiring a face image in first information and second information; extracting 72 human face features according to the Haar features; and weighting by using an AHP algorithm according to the importance of each face feature, and finally classifying and identifying by using an SVM vector machine to realize the matching between the first information and the second information. Of course, there are many other different choices for the matching method of the person information for those skilled in the art, and there are no particular limitations here.

It should be noted that, first, the first item information can be automatically obtained by comparing the first information with the second information, so that data loss of the first item information due to human factors is avoided, and the performability of the anti-theft method is improved. Secondly, the above scheme only describes the general situation, namely the situation that a user places an article from a corridor and then enters a residential building through the corridor, and other special situations, such as the situation that the user enters the corridor from the residential building, places the article and then leaves from an entrance of the corridor; a user enters from the entrance of the corridor, places an article and then leaves from the entrance of the corridor; the user enters the corridor from the residential building, places articles and returns to the residential building again. The information acquisition methods for these special cases are the same as the information acquisition methods described above, and therefore are not described herein again.

S200, calculating to obtain a first feature vector of the first article information.

Specifically, the first feature vector includes pattern features, contour features, and other features of the first item information. The pattern features include, but are not limited to, pattern information in the article, and may also include other security features, such as colors, letters, shapes, materials, and the like. The contour features include changes in the outer contour curve of the first item information and the size of the area of the image. The contour feature acquisition method adopts Hu invariant moment of a description method of image walking of an area and utilizes 7 invariant moments constructed by second-order and third-order normalized central moments. Other features include other more prominent features such as license plate features, RFID features. However, in the actual processing process, the license plate features are difficult to completely capture due to the problem of shooting angles; the prevalence of RFID chips is relatively low. Other features are more or less problematic and will therefore not be described further here.

The pattern feature acquisition method comprises the following steps:

s211, converting the first article information into a preset angle through a shooting angle perspective principle;

the planar projective transformation is:

wherein, F_wa(s, t, u, c) is light field information of the picture after plane projection transformation; f_in((s, T, u, v) is light field information present in the first information; T_dA spatial transformation matrix for the d picture; d is the number of the plane projection transformation pictures;

s212, constructing a Hessian matrix, and extracting N characteristic patterns according to Haar characteristics;

s213, counting 4 values of the sum of horizontal direction values sigma dx, the sum of vertical direction values sigma dy, the sum of horizontal direction absolute values sigma dx and the sum of vertical direction absolute values sigma dy in each characteristic pattern;

s213, taking the 4 values as a characteristic vector of the characteristic pattern;

s214, combining the characteristic vectors of each characteristic pattern to obtain a pattern characteristic vector; the feature points of the feature vector of the pattern feature are 4N;

s300, third information is obtained, first processing is carried out on the third information, and second item information and second personnel information are obtained.

Specifically, the third information is video and/or image information of a first area shot by the monitoring device; the system is mainly used for collecting video and/or image information of people and objects leaving a corridor from an entrance; the first area is an entrance of a corridor; the shooting angles of the monitoring device of the third information and the monitoring device of the first information are in mirror symmetry. The second article information is the characteristic information of the second article; the second article information is an article leaving the corridor from the entrance; the second person information is face information of the second person; the second person is a person appearing in the third information at the same time as the second article, and the boundary between the second person information and the second article information is overlapped.

The first processing method comprises the following steps:

and S301, preprocessing the third information to obtain an invalid area. Specifically, the method for acquiring the invalid region includes: acquiring pixel points in the image information of the third information; calculating the pixel difference value of the pixel points at the same position in the adjacent two frames of image information; judging whether the pixel difference value is smaller than a preset value, if so, taking the pixel point as a background point; and accumulating the background points to obtain an invalid area.

Step S302, performing mask processing on the invalid area;

step S303, dividing the image in the third information at the later stage of mask processing into a plurality of super-pixel color areas based on a super-pixel division SLIC method to form a pixel matrix, and calculating the color and brightness mean value of each element in the pixel matrix; comparing the color value and the brightness between two adjacent elements to obtain a similarity measurement; wherein the similarity measure between the two elements is:

wherein, S (B)_i,B_j) A similarity measure between the two elements; b is_i，B_jAre respectively provided withRepresenting a coloured area represented by two elements, S_c(B_i,B_j) Representing a color feature metric between two elements; s_r(B_i,B_j) Representing a measure of the luminance characteristic between two elements, S_l(B_i,B_j) Representing a position measure between two elements; and a and b are respectively the weight of the color characteristic and the brightness characteristic, the sum of the weight of the color characteristic and the brightness characteristic is 1, and c is a correction parameter with the value range of 0-10.

If the similarity measurement is smaller than the elements in the preset range, forming a second pixel matrix, and finally combining the second pixel matrix into a region; then clustering images of the color images by using LAB information through a python algorithm after super-pixel equalization processing, and reducing the gray value of a class in contact with the image boundary in a clustering result to 0; and finally obtaining the second personnel information and the second article information.

S400, performing first analysis on the second item information and the first item information to obtain a first analysis result.

Specifically, the first analysis determines whether the third information includes the first item information, if so, step S500 is executed, and if not, the third information is obtained again.

The first analysis process specifically comprises the following steps:

s410, calculating a second feature vector of the second item information; the second feature vector comprises color features, contour features and special features of the article information in the first information. The specific method can be seen in step S200

And S420, calculating the similarity value of the first feature vector and the second feature vector.

Specifically, the contour features and the pattern features are fused to form a 4N-dimensional vector matrix, and then the first feature vector and the second feature are calculated according to Euclidean distance, wherein the calculation method of the similarity value of the first feature vector and the second feature comprises the following steps:

wherein P is the similarity value of the first feature vector and the second feature, L_ijThe characteristic elements of the ith row and the jth column of the 4N-dimensional vector matrix formed by the first characteristic vector are expressed; said L_ij' denotes the feature elements of the ith row and the jth column of the 4N-dimensional vector matrix formed by the second feature vector.

S430, judging that the similarity meets the threshold requirement; if yes, whether the first article information appears in the first information or not is indicated.

By the method of combining multiple characteristics, the calculation complexity is reduced while the higher accuracy is ensured, and the corresponding speed of the method is improved.

S500, judging whether to execute a second analysis according to the first analysis result to obtain a second analysis result; the second analysis is to judge whether the second person information belongs to first person information.

Specifically, it is determined whether the second person is the owner of the first item or has a relationship of relativity, friendship, and roommates with the owner by determining whether the second person information belongs to the first person information. The specific judgment method can match the first person information with the second person information. The matching method of the personnel information comprises the following steps: firstly, acquiring a face image in first information and second information; extracting 72 human face features according to the Haar features; and weighting by using an AHP algorithm according to the importance of each face feature, and finally classifying and identifying by using an SVM vector machine to realize the matching between the first information and the second information. Of course, there are many other different choices for the matching method of the person information for those skilled in the art, and there are no particular limitations here.

S600, judging whether to execute a first instruction according to the second analysis result; the first instruction includes an alert prompt.

Specifically, if the second person information does not belong to the first person information, it can be determined that the second person is a dangerous person; executing a first instruction, wherein the first instruction comprises but is not limited to an alarm prompt, and the alarm prompt is specifically used for transmitting alarm information to a guard system and prompting a guard to make relevant confirmation; and a certain channel can be closed, and the second person is required to carry out relevant information confirmation.

Example 2

Based on the same inventive concept as the community corridor anti-theft method in the foregoing embodiment 1, the present invention also provides a corridor anti-theft device, as shown in fig. 2, the device includes:

afirst acquisition unit 11 configured to acquire first item information and first person information; the first person information has a first association with the first item information;

thefirst processing unit 12 is configured to calculate a first feature vector of the first item information;

the second obtainingunit 13 is configured to obtain third information, perform first processing on the first information, and obtain second item information and second person information;

thesecond processing unit 14 is configured to perform a first analysis on the second item information and the first item information to obtain a first analysis result;

afirst judging unit 15, configured to judge whether to perform a second analysis according to the first analysis result, so as to obtain a second analysis result; the second analysis is to judge whether the second person information belongs to first person information;

afirst execution unit 16, configured to determine whether to execute a first instruction according to the second analysis result; the first instruction includes an alert prompt.

The device, still include:

a third acquisition unit configured to acquire first information of the first area; the first area is an entrance of a corridor; the first information is video and/or image information of a first area shot by the monitoring device;

a fourth acquiring unit configured to acquire second information of the second area; the second area is an exit of a corridor, and the first information is video and/or image information of the first area shot by the monitoring device;

the third processing unit is used for respectively preprocessing the first information and the second information; the preprocessing is to perform mask processing on invalid areas of the first information and the second information;

the fourth processing unit is used for respectively segmenting the first information and the second information to obtain article information and personnel information of the first information and the second information;

the second judgment unit is used for matching the personnel information of the first information and the second information within preset time and judging whether the personnel information is the same target object;

and the fifth processing unit is used for calculating the missing article information in the second information to obtain the first article information.

The device, still include:

a fifth obtaining unit, configured to obtain a pixel point in the image information of the first information and/or the second information, and obtain a pixel value of the pixel point;

the sixth processing unit is used for calculating the pixel difference value of the pixel point at the same position in the adjacent two frames of image information;

the third judging unit is used for judging whether the pixel difference value is smaller than a preset value or not, and if so, the pixel point is a background point;

and the seventh processing unit is used for accumulating the background points to obtain an invalid area.

The device, still include:

the eighth processing unit is used for converting the first article information into a preset angle according to a shooting angle perspective principle;

the ninth processing unit is used for constructing a Hessian matrix and extracting N characteristic patterns according to Haar characteristics;

a tenth processing unit, configured to count 4 values of a sum of horizontal direction values, a sum of vertical direction values, a sum of horizontal direction absolute values, and a sum of vertical direction absolute values in each feature pattern; taking the 4 values as a feature vector of the feature pattern;

the eleventh processing unit is used for merging the characteristic vectors of each characteristic pattern to obtain a pattern characteristic vector; the feature points of the feature vector of the pattern feature are 4N.

The device, still include:

a twelfth processing unit, configured to perform preprocessing on the third information to obtain an invalid region;

a thirteenth processing unit configured to perform mask processing on the invalid region;

a fourteenth processing unit, configured to divide the image in the third information at the later stage of the mask processing into a plurality of super-pixel color regions based on a super-pixel division SLIC method, so as to form a pixel matrix;

a fifteenth processing unit, configured to calculate a color and luminance mean value of each element in the pixel matrix;

the fourth judging unit is used for comparing the color value and the brightness between two adjacent elements to obtain the similarity measurement;

a sixteenth processing unit, configured to form a second pixel matrix if the similarity measure is smaller than the element in the predetermined range, and finally merge the second pixel matrix into an area;

a seventeenth processing unit, configured to cluster the color images by using LAB information according to the python algorithm after the super-pixel equalization processing, and reduce a gray value in a class in contact with an image boundary in a clustering result to 0; and obtaining the second personnel information and the second article information.

The device, still include:

the eighteenth processing unit calculates a second feature vector of the second item information;

a nineteenth processing unit that calculates a similarity value of the first feature vector and the second feature vector;

a fifth judging unit, configured to judge that the similarity satisfies a threshold requirement; if yes, whether the first article information appears in the first information or not is indicated.

Various changes and specific examples of a method for preventing a corridor from being stolen in a community in the foregoing embodiment 1 are also applicable to an electric vehicle identification device in this embodiment, and a method for implementing a corridor anti-theft device in this embodiment is clearly known to those skilled in the art from the foregoing detailed description of a method for preventing a corridor from being stolen, so for the sake of brevity of the description, detailed description is omitted here.

Example 3

Based on the same inventive concept as one of the above-mentioned embodiments of the anti-theft method for the corridor of a cell, the present invention further provides an anti-theft server for the corridor, as shown in fig. 3, fig. 3 is an exemplary electronic device in embodiment 3, and includes amemory 304, aprocessor 302, and a computer program stored on thememory 304 and executable on theprocessor 302, and when theprocessor 302 executes the program, the steps of any one of the above-mentioned methods for remote vital sign monitoring are implemented.

Where in fig. 3 a bus architecture (represented by bus 300),bus 300 may include any number of interconnected buses and bridges,bus 300 linking together various circuits including one or more processors, represented byprocessor 302, and memory, represented bymemory 304. Thebus 300 may also link together various other circuits such as peripherals, voltage regulators, power management circuits, and the like, which are well known in the art, and therefore, will not be described any further herein. Abus interface 305 provides an interface between thebus 300 and thereceiver 301 andtransmitter 303. Thereceiver 301 and thetransmitter 303 may be the same element, i.e., a transceiver, providing a means for communicating with various other apparatus over a transmission medium.

Theprocessor 302 is responsible for managing thebus 300 and general processing, and thememory 304 may be used for storing data used by theprocessor 302 in performing operations.

Example 4

Based on the same inventive concept as the anti-theft method for the residential corridor in the foregoing embodiments, the present invention also provides a computer-readable storage medium having stored thereon a computer program which, when executed by a processor, implements the steps of: acquiring first article information and first person information; the first person information has a first association with the first item information; calculating to obtain a first feature vector of the first article information; acquiring third information, and performing first processing on the first information to obtain second item information and second personnel information; performing first analysis on the second article information and the first article information to obtain a first analysis result; judging whether to execute a second analysis according to the first analysis result to obtain a second analysis result; the second analysis is to judge whether the second person information belongs to first person information; judging whether to execute a first instruction according to the second analysis result; the first instruction includes an alert prompt.

One or more technical solutions in the embodiments of the present invention at least have one or more of the following technical effects: objects stored in the corridor are taken as main bodies, so that the man-made interference is reduced, the judgment of uncontrollable factors on the method is reduced, and the misjudgment rate is reduced; the first information and the second information are compared, the first article information can be automatically acquired, data loss of the first article information caused by human factors is avoided, and the performability of the anti-theft method is improved. Through improving from the perspective of vision, the pixel points on the boundary are analyzed, the personnel information and the article information are accurately segmented, and shadow interference caused by shadows or other light reasons can be effectively eliminated.

It should be noted that the various features described in the above embodiments may be combined in any suitable manner without departing from the scope of the invention. The invention is not described in detail in order to avoid unnecessary repetition.

Claims

1. A method for theft prevention for a residential corridor, the method comprising:

acquiring first article information and first person information; the first person information has a first association with the first item information;

calculating to obtain a first feature vector of the first article information;

acquiring third information, and performing first processing on the first information to obtain second item information and second personnel information;

performing first analysis on the second article information and the first article information to obtain a first analysis result;

judging whether to execute a second analysis according to the first analysis result to obtain a second analysis result; the second analysis is to judge whether the second person information belongs to first person information;

judging whether to execute a first instruction according to the second analysis result; the first instruction includes an alert prompt.

2. The method of claim 1, further comprising:

acquiring first information of a first area; the first area is an entrance of a corridor; the first information is video and/or image information of a first area shot by the monitoring device;

acquiring second information of a second area; the second area is an exit of a corridor, and the first information is video and/or image information of the first area shot by the monitoring device;

respectively preprocessing the first information and the second information; the preprocessing is to perform mask processing on invalid areas of the first information and the second information;

respectively segmenting the first information and the second information to obtain article information and personnel information of the first information and the second information;

matching the personnel information of the first information and the second information within preset time, and judging whether the personnel information is the same target object;

if yes, calculating the article information which is lacked in the article information in the second information to obtain first article information.

3. The method of claim 2, further comprising:

acquiring a pixel point in the image information of the first information and/or the second information, and acquiring a pixel value of the pixel point;

calculating the pixel difference value of the pixel points at the same position in the adjacent two frames of image information;

judging whether the pixel difference value is smaller than a preset value, if so, taking the pixel point as a background point;

and accumulating the background points to obtain an invalid area.

4. The cell corridor anti-theft method according to claim 1, wherein the first feature vector includes pattern features, contour features, and other features of the first item information.

5. The method of claim 4, further comprising:

converting the first article information into a preset angle through a shooting angle perspective principle;

constructing a Hessian matrix, and extracting N characteristic patterns according to Haar characteristics;

counting 4 values of the sum of the horizontal direction values, the sum of the vertical direction values, the sum of the horizontal direction absolute values and the sum of the vertical direction absolute values in each characteristic pattern;

taking the 4 values as a feature vector of the feature pattern;

combining the characteristic vectors of each characteristic pattern to obtain a pattern characteristic vector; the feature points of the feature vector of the pattern feature are 4N.

6. The method of claim 1, further comprising:

preprocessing the third information to obtain an invalid area;

masking the invalid region;

dividing an image in the third information at the later stage of mask processing into a plurality of super-pixel color areas on the basis of a super-pixel segmentation SLIC method to form a pixel matrix;

calculating the color and brightness mean value of each element in the pixel matrix;

comparing the color value and the brightness between two adjacent elements to obtain a similarity measurement;

if the similarity measurement is smaller than the elements in the preset range, forming a second pixel matrix, and finally combining the second pixel matrix into a region;

clustering images of the color images by using LAB information through a python algorithm after super-pixel equalization processing, and reducing the gray value of a class contacting with the image boundary in a clustering result to 0;

and obtaining the second personnel information and the second article information.

7. The method of claim 1, further comprising:

calculating a second feature vector of the second article information;

calculating the similarity value of the first feature vector and the second feature vector;

judging whether the similarity meets the threshold requirement; if yes, whether the first article information appears in the first information or not is indicated.

8. An anti-theft device for corridors, characterized in that it comprises:

9. A server for passageway theft protection, comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor implements the steps of the method according to any one of claims 1 to 7 when executing the program.

10. A computer-readable storage medium having computer program instructions stored thereon which, when executed by a processor, implement the steps of the method of any one of claims 1-7.