CN111104554A

CN111104554A - Automatic door control method and system based on contour detection and head and shoulder detection

Info

Publication number: CN111104554A
Application number: CN201911336041.7A
Authority: CN
Inventors: 杨帆; 袁顺; 刘凯; 徐善俊
Original assignee: Nanjing Zhenshi Intelligent Technology Co Ltd
Current assignee: Xiaoshi Technology Jiangsu Co ltd
Priority date: 2019-12-23
Filing date: 2019-12-23
Publication date: 2020-05-05
Anticipated expiration: 2039-12-23
Also published as: CN111104554B

Abstract

The invention provides an automatic door control method and system based on contour detection and head and shoulder detection, which are characterized in that non-human objects and animals except a detection area triggered by an automatic door opening signal are judged and excluded through contour detection, the head and shoulder in a human-shaped contour are identified through head and shoulder detection, the head and shoulder displacement in a video frame is calculated to judge whether a real person exists, and further whether a person has a tendency of needing to trigger a door opening signal is judged through tracking the position of a coordinate point by the head and shoulder, so that the defects that the object and the animal can trigger the automatic door and the situation that the person does not wish to mistakenly trigger the door opening through the detection area are solved. Meanwhile, the head and shoulder detection is used for judging whether the virtual area has a store-exiting tendency tracking ID and a historical sending signal in real time to judge a signal needing to be sent next step. As long as there is a tendency of going out of a store to track ID in the detection area of opening the door, the automatically-controlled door keeps the open state, effectively solve the problem that the opening time is not enough or too long when the automatically-controlled door is closed in a delayed way.

Description

Automatic door control method and system based on contour detection and head and shoulder detection

Technical Field

The invention relates to the technical field of head and shoulder detection, in particular to an automatic door control method based on contour detection and head and shoulder detection.

Background

Automatic door among the prior art, automatic door of hotel, museum's automatic door or other open public places's automatic door carry out induction control through infrared switch usually, when the object is close, open the door action through infrared sensor detection, close the door after the object leaves the detection zone. However, in practical application, it is not the object, animal or real person that infrared pulse detection control automatic door can't discern triggering the signal of opening the door, for example, customer carries the pet to visit under the market shop scene, if there is the possibility that the signal of opening the door is lost to false triggering under the non-traction state of pet, so lead to carrying on the flow in business circle/market and the collection and the intelligent analysis in-process noise more, can take place great deviation, lead to the analysis result not to have realistic meaning.

In real life, the store/hotel entrance often has a exhibition situation, and if an object exists in an infrared detection area or at the edge, a door opening signal is frequently triggered by using the traditional infrared pulse detection. Moreover, the automatic door which is opened by using a single signal triggering and is closed in a delayed mode can only be opened for a fixed time, and the passing speed is insufficient, so that the automatic door cannot pass through or is clamped.

Disclosure of Invention

The invention aims to provide an automatic door control method based on contour detection and head and shoulder detection, which is used for rapidly identifying through contour detection and head and shoulder detection and independently controlling an automatic door through door opening information or door closing information.

In order to achieve the above object, the present invention provides an automatic door control method based on contour detection and head and shoulder detection, comprising the following steps:

step 1, presetting a virtual door in a video picture acquired by a camera corresponding to the automatic door, recording vertex coordinates of the virtual door, presetting a rectangular frame outside the virtual door as a virtual detection area, and recording the vertex coordinates of the rectangular frame;

step 2, decoding the video stream acquired by the camera in real time to obtain a video frame;

step 3, carrying out contour detection on the video frame, extracting the human-shaped contour in the video frame, generating a tracking ID, recording the vertex coordinate point of the human-shaped contour, and then outputting the video frame with the tracking ID and the tracking ID of each human-shaped contour in the video frame, namely contour TrackID, and contour vertex coordinates;

step 4, performing head and shoulder detection on the video frame with the tracking ID output in the step 3, extracting each head and shoulder contour in the video frame, generating the tracking ID, namely the head and shoulder TrackID, and recording the head and shoulder vertex coordinates; then comparing the contour vertex coordinates with the head shoulder vertex coordinates, if the comparison is consistent, updating the head shoulder TrackID to be the contour TrackID, otherwise, not updating, and outputting the head shoulder contour, the head shoulder TrackID and the head shoulder vertex coordinates;

step 5, according to the head shoulder contour, the head shoulder TrackID and the head shoulder vertex coordinates output in the step 4, detecting whether the head shoulder TrackID is a new TrackID in a TrackID monitoring list, if the head shoulder TrackID is the new TrackID, inserting the head shoulder TrackID monitoring list into the TrackID monitoring list and updating the TrackID monitoring list, recording the head shoulder vertex coordinates corresponding to the head shoulder TrackID as a starting coordinate point, and if the head shoulder TrackID is not the new TrackID, indicating the existing TrackID and recording the head shoulder vertex coordinates;

step 6, comparing the head and shoulder vertex coordinates with the coordinates of the detection area: if the head and shoulder vertex coordinates are outside the virtual detection area, no processing is performed, and if the head and shoulder vertex coordinates are inside the virtual detection area, the step 7 is performed;

step 7, outputting the head shoulder TrackID and entering step 8 in response to the going-out tendency according to the trend of the head shoulder TrackID in the virtual detection area, otherwise, discarding and not acting;

step 8, judging whether the virtual door is in a closed state: if the door is in a closed state, sending a door opening signal to the door access controller to control the opening of the automatic door; if the virtual door is in the open state, no processing is performed.

Further preferably, in the step 8, the method further includes:

and detecting the TrackID monitoring list, if the head shoulder TrackID still exists, not operating, and if the head shoulder TrackID does not exist, sending a door closing signal to the door access controller to control closing of the automatic door.

Further preferably, the specific operation of step 7 includes:

marking the last sitting mark of the head shoulder TrackID before entering the virtual detection area as point A, marking the first point of the head shoulder TrackID after entering the virtual detection area as point B, connecting the point A to the point B as an extension line to calculate the intersection point with the virtual door, judging that the door-out tendency exists if the intersection point exists, outputting the head shoulder TrackID, and discarding the tracking if the intersection point does not exist.

Further preferably, a second point after the head shoulder TrackID enters the virtual detection area is marked as a point C, the point B is connected to the point C to form an extension line, the intersection points with the virtual door are calculated, if the intersection points of A-B and B-C exist, the door-out tendency is judged, a pre-door-opening signal is sent, the head shoulder TrackID judged to send the door-opening signal is output, and if the intersection points do not exist, the tracking is discarded.

According to the disclosure of the present invention, there is also provided an automatic door control system based on contour detection and head and shoulder detection, comprising:

a module for presetting a virtual door in a video picture collected by a camera corresponding to the automatic door, recording vertex coordinates of the virtual door, presetting a rectangular frame outside the virtual door as a virtual detection area, and recording the vertex coordinates of the rectangular frame;

a module for decoding the video stream collected by the camera in real time to obtain a video frame;

a module for detecting the outline of the video frame, extracting the human-shaped outline in the video frame, generating a tracking ID, recording the vertex coordinate point of the human-shaped outline, and then outputting the video frame with the tracking ID and the tracking ID of each human-shaped outline in the video frame, namely the outline TrackID, and the outline vertex coordinates;

the tracking ID generation device is used for carrying out head and shoulder detection on the output video frame with the tracking ID, extracting each head and shoulder contour in the video frame, generating the tracking ID, namely the head and shoulder TrackID, and recording head and shoulder vertex coordinates; then comparing the contour vertex coordinates with the head shoulder vertex coordinates, if the comparison is consistent, updating the head shoulder TrackID to be the contour TrackID, otherwise, not updating, and outputting the head shoulder contour, the head shoulder TrackID and the head shoulder vertex coordinates;

a module for detecting whether the head shoulder TrackID is a new TrackID in a TrackID monitoring list according to the output head shoulder contour, the head shoulder TrackID and the head shoulder vertex coordinates, if the head shoulder TrackID is the new TrackID, inserting the head shoulder TrackID monitoring list into the TrackID monitoring list and updating the TrackID monitoring list, recording the head shoulder vertex coordinates corresponding to the head shoulder TrackID as a starting coordinate point, and if the head shoulder TrackID is not the new TrackID, indicating the existing TrackID and recording the head shoulder vertex coordinates;

a module for comparing the head and shoulder vertex coordinates with the coordinates of the detection area, wherein if the head and shoulder vertex coordinates are outside the virtual detection area, no processing is performed, and if the head and shoulder vertex coordinates are inside the virtual detection area, the tendency to go out is judged;

a module for outputting the TrackID of the head shoulder according to the trend of the TrackID of the head shoulder in the virtual detection area and responding to the tendency of going out, otherwise, discarding the head shoulder TrackID which does not act;

a module for determining whether the virtual door is in a closed state, wherein: if the door is in a closed state, sending a door opening signal to the door access controller to control the opening of the automatic door; if the virtual door is in the open state, no processing is performed.

It should be understood that all combinations of the foregoing concepts and additional concepts described in greater detail below can be considered as part of the inventive subject matter of this disclosure unless such concepts are mutually inconsistent. In addition, all combinations of claimed subject matter are considered a part of the presently disclosed subject matter.

The foregoing and other aspects, embodiments and features of the present teachings can be more fully understood from the following description taken in conjunction with the accompanying drawings. Additional aspects of the present invention, such as features and/or advantages of exemplary embodiments, will be apparent from the description which follows, or may be learned by practice of specific embodiments in accordance with the teachings of the present invention.

Drawings

The drawings are not intended to be drawn to scale. In the drawings, each identical or nearly identical component that is illustrated in various figures may be represented by a like numeral. For purposes of clarity, not every component may be labeled in every drawing. Embodiments of various aspects of the present invention will now be described, by way of example, with reference to the accompanying drawings, in which:

FIG. 1 is an exemplary flow chart of the method of automatic door control of the present invention.

Fig. 2 is a logical example of whether or not the track point direction of the head shoulder TrackID satisfies the out-of-store tendency condition.

Fig. 3A and 3B are two specific examples of whether or not the track point direction of the head shoulder TrackID satisfies the out-of-store tendency condition.

Fig. 4 is still another specific example of whether or not the track point direction of the head shoulder TrackID satisfies the out-of-store tendency condition.

Detailed Description

In order to better understand the technical content of the present invention, specific embodiments are described below with reference to the accompanying drawings.

In this disclosure, aspects of the present invention are described with reference to the accompanying drawings, in which a number of illustrative embodiments are shown. Embodiments of the present disclosure are not necessarily intended to include all aspects of the invention. It should be appreciated that the various concepts and embodiments described above, as well as those described in greater detail below, may be implemented in any of numerous ways, as the disclosed concepts and embodiments are not limited to any one implementation. In addition, some aspects of the present disclosure may be used alone, or in any suitable combination with other aspects of the present disclosure.

In combination with the drawings, the invention aims to provide an automatic door control method based on contour detection and head and shoulder detection, which realizes the false detection judgment of personnel based on contour detection, head and shoulder detection and coordinate point displacement judgment, records the tracking coordinate points and the judgment of the position of the automatic door based on the head and shoulder detection, judges whether the head and shoulder detection effectively tracks the detection ID in the detection area or not and judges the next action according to historical data, and controls the opening and closing of the automatic door independently through door opening and closing information so as to effectively prevent the automatic door from falsely detecting non-human objects, animals and non-exit intention of personnel to trigger an opening signal and prevent the problem of being clamped due to the closing of the automatic door (for example, the traditional automatic door control can be clamped) during passing.

Referring to fig. 1-4, the method for controlling an automatic door based on contour detection and head and shoulder detection according to the present invention comprises the following steps:

step 1, presetting a virtual door in a video picture acquired by a camera corresponding to the automatic door, as shown in fig. 2, recording vertex coordinates of the virtual door, presetting a rectangular frame outside the virtual door as a virtual detection area, and recording the vertex coordinates of the rectangular frame;

With reference to fig. 1 and 2, in step 8, the method further includes:

Preferably, as shown in fig. 2, 3A and 3B, the specific operation of step 7 includes:

Preferably, in the implementation of the present invention, the automatic door is controlled to maintain the open state in response to the presence of a tracking ID that tends to go out, i.e., a head shoulder TrackID, within the virtual detection area.

As shown in fig. 2 and 3A-3B, the tendency of going out is exemplarily shown.

In the figure, the video of the camera is the exit area of the automatic door in the hotel, in the video area of the camera, a straight line ab is drawn at the position of the automatic door, and a rectangular frame is drawn behind the door. The TrackID is detected out of trigger logic judgment in the rectangular frame, and an intersection point X of a connecting line from a TrackID coordinate point A to a coordinate point B and a straight line ab is calculated. If X points exist, the judgment is true, and if X points do not exist, the judgment is false.

In another embodiment, as shown in fig. 4, the second point after the head shoulder TrackID enters the virtual detection area is marked as point C, the point B is connected to point C to form an extension line, the intersection points with the virtual door are calculated, if both a-B and B-C exist the intersection points with the virtual door (point X, Y), the door tendency is judged to exist, the pre-opening signal is sent, the head shoulder TrackID judged to send the door opening signal is output, and if no intersection point exists, the tracking is discarded. Thus, the accuracy of the determination is further enhanced.

With reference to fig. 1, in the automatic door control scheme of the present invention, in the system configuration, a conventional camera may be adopted, or an original camera (generally, a camera with an infrared imaging function) of the door control system may be deployed, and a software program compiled according to the method is deployed in a system background and deployed in a server or a server system, so as to process a video stream acquired by the camera in real time, and perform door opening and closing control based on the above-mentioned contour and head and shoulder detection after obtaining a video frame.

Based on video image acquisition and contour head-shoulder processing, a recognition algorithm with small volume operation is preferentially adopted, for example, the recognition algorithm based on Opencv has high detection and recognition efficiency which can reach millisecond level, real-time ID tracking and real-time door opening and closing control are realized during door opening and closing, and compared with the traditional infrared detection, the invention can fundamentally eliminate the problem that a non-human object triggers a door opening signal through a contour detection technology; particularly, under the condition that the exhibition is arranged at the entrance of the hotel, the traditional infrared pulse detection is used for triggering the door opening signal frequently, and the problem that the door opening signal is triggered by the fixed object can be solved by combining the head and shoulder tracking algorithm and the virtual area judgment.

It should be noted that the contour detection used in the present invention refers to a process of extracting a contour of an object by processing a background, a texture inside the object, and noise interference in a digital image including the object and the background. The contour Detection mentioned in the invention can adopt Human contour Detection (Real-Time Human Detection using contour currents) based on contour clues or Human contour recognition algorithm based on Opencv, and the contour Detection can be packaged into SDK, namely, a video frame comprising Human contours is used as input, and tracking IDs of all Human contours in the video frame and coordinate positions of the Human contours are output. A person in a video frame corresponds to a human-shaped outline.

The head and shoulder detection technology is a process of recognizing an input video frame image based on an image processing technology, detecting a head and shoulder in the video frame, and outputting a tracking ID of the head and shoulder and a coordinate position of the head and shoulder in the video frame. The head and shoulder detection mentioned in the invention adopts a head and shoulder detection algorithm based on feature description or classifier learning, and is packaged into SDK, namely, the video frame is used as input, and the tracking IDs of all the heads and shoulders in the video frame and the coordinate positions of the heads and shoulders are output. A person in a video frame corresponds to a head and a shoulder.

Therefore, the invention eliminates non-human objects and animals except the detection area triggered by the automatic door opening signal through the contour detection, identifies the head and the shoulder in the human-shaped contour through the head and shoulder detection, calculates the displacement of the head and the shoulder in the video frame and judges whether the human is real or not, and reduces the false detection probability and the tracking loss probability. And further judging whether the personnel have a tendency of triggering the door opening signal or not by tracking the position of the coordinate point through the head and the shoulder. The method solves the defect that objects and animals can trigger the automatic door and the situation that people do not wish to go out of a store mistakenly trigger the door to open through a detection area.

And judging whether the virtual area has a store-exiting tendency tracking ID and a historical transmission signal in real time through head and shoulder detection to judge that the signal needs to be transmitted next step. As long as there is a tendency of going out of a store to track ID in the detection area of opening the door, the automatically-controlled door keeps the open state, effectively solve the problem that the opening time is not enough or too long when the automatically-controlled door is closed in a delayed way.

For example, scene one, pet false triggering door opening information scene

The customer pulls the pet to the store, pays attention to talk with reception staff when handling the check-in procedure, and needs to make actions such as showing identity documents, and the pet carried with him is easy to keep away. The traditional hotel gate uses infrared pulse detection, and the hotel gate can be automatically opened when a pet appears in a detection area. The risk of injury or injury to others for an uncontrolled pet is unpredictable. According to the invention, the door opening signal cannot be triggered when the pet appears in the detection area by using a technology based on human-shaped contour detection, so that the problem that the pet is lost in a non-traction state is solved.

Scene two, hotel entrance exhibition scene

Due to the large traffic, hotel entrances are often occupied by commercial marketing deployments. Hotel doors using traditional infrared pulse detection are repeatedly triggered to open the door by marketing personnel who statically deploy materials and wander. In the scene, the door opening triggering information is triggered by infrared pulse detection, so that the defect of serious false detection exists, the door opening signal is triggered only by judging the human shape with the motion gesture by combining head and shoulder tracking of the human shape profile, static materials do not have the human shape profile and cannot be detected (the prerequisite condition of no detection), and the door opening signal cannot be triggered even if the human shape profile has no motion gesture, so that the problem of false triggering of the static materials is solved. The movement of the loitering marketing personnel has certain range characteristics, the virtual area is set to track the position and the movement direction of the personnel in the video area, and the situation that the personnel moves into the virtual area after being detected outside the virtual area and the direction points to a hotel gate can be judged to have a tendency of going out to trigger a door opening signal, so that the problem that the loitering personnel continuously trigger the door opening signal in a small range is solved.

Scene three, person crowding overlapping scene

In a video, if people are crowded and overlapped, the problem that the complete human figure outline cannot be extracted and tracking loss is easy to occur when the traditional human figure detection is used for controlling the opening and closing of a door is solved.

The present invention can also be configured as follows with reference to the embodiments of the present invention shown in the drawings and described above.

An automatic door control system based on contour detection and head and shoulder detection, comprising:

Preferably, the method further comprises the following steps:

and the module is used for detecting the TrackID monitoring list, if the head shoulder TrackID still exists, the operation is not carried out, and if the head shoulder TrackID does not exist, a door closing signal is sent to the access controller to control the closing of the automatic door.

Preferably, the trend is set to be determined by:

Although the present invention has been described with reference to the preferred embodiments, it is not intended to be limited thereto. Those skilled in the art can make various changes and modifications without departing from the spirit and scope of the invention. Therefore, the protection scope of the present invention should be determined by the appended claims.

Claims

1. An automatic door control method based on contour detection and head and shoulder detection is characterized by comprising the following steps:

2. The method for automatic door control based on contour detection and head and shoulder detection as claimed in claim 1, wherein in the step 8, further comprising:

3. The method for automatic door control based on contour detection and head and shoulder detection as claimed in claim 1, wherein the specific operation of step 7 comprises:

4. The automatic door control method based on contour detection and head shoulder detection as claimed in claim 3, characterized in that the second point after the head shoulder TrackID enters the virtual detection area is marked as point C, the point B is connected to the point C to be an extension line, the intersection points with the virtual door are calculated, if the intersection points of A-B and B-C with the virtual door exist, the door-out tendency is judged to exist, the pre-door-opening signal is sent, the head shoulder TrackID judged to send out the door-opening signal is output, and if the intersection points do not exist, the tracking is discarded.

5. The method for automatic door control based on contour detection and head and shoulder detection as claimed in any one of claims 1-4, wherein the automatic door is controlled to keep open in response to the tracking ID of the door exit tendency, namely the head and shoulder TrackID, existing in the virtual detection area.

6. The method for automatic door control based on contour Detection and head-shoulder Detection as claimed in claim 1, wherein the contour Detection adopts Human contour Detection based on contour cue (Real-Time Human Detection using contour currents) or Human contour recognition algorithm based on Opencv.

7. The method for automatic door control based on contour detection and head and shoulder detection as claimed in claim 1, wherein the head and shoulder detection adopts a head and shoulder detection algorithm based on feature description or classifier learning.

8. An automatic door control system based on contour detection and head and shoulder detection, comprising:

9. The system of claim 8, further comprising:

10. The system of claim 8, wherein the trend is configured to be determined by: