Security and protection system for analyzing distortionless image acquired based on ball-based camera

Abstract

The invention discloses a security and protection system for analyzing distortionless images acquired based on a ball-based camera, which belongs to the technical field of security and protection and comprises an initialization module, an acquisition module, a preprocessing module, a target detection module, a size estimation module, a color estimation module, a data storage module and a security and protection module. Video frames are collected through a camera, a rider and a bicycle in the video frames are detected through a target detection module, the sizes and colors of the rider and the bicycle in the video frames are estimated through a size estimation module and a color estimation module, and a security program is automatically triggered when the rider is separated from the bicycle. The size and color similarity between a rider recombined with the bicycle and a previous rider is compared through the security module, whether the rider and the previous rider are the same rider or not is judged, whether an alarm is given or not is judged, compared with the prior art, a new implementation scheme can be provided for security, and meanwhile the security accuracy rate is high.

Description

Technical Field

[0001] The present invention belongs to the technical field of security, and in particular relates to a security system for collecting distortion-free images for analysis based on a spherical camera. Background Art

[0002] A security system is an electronic system or network composed of security products and related technologies for the purpose of maintaining public safety.

[0003] The current security system mainly focuses on face recognition. If the face is blocked or difficult to identify during the security process, it will affect the security work.

[0004] In view of this, a security system based on a spherical camera to collect distortion-free images for analysis is designed to solve the above problems. Summary of the invention

[0005] In order to solve the problems raised in the above background technology, the present invention provides a security system based on a spherical camera to collect distortion-free images for analysis, which has the characteristics of being able to provide a new implementation plan for security and having a high security accuracy.

[0006] To achieve the above object, the present invention provides the following technical solution: a security system based on a spherical camera collecting distortion-free images for analysis, comprising:

[0007] Initialization module, camera parameter initialization;

[0008] Acquisition module, the camera acquires video frames;

[0009] A preprocessing module preprocesses the collected video frames;

[0010] The object detection module inputs the preprocessed video frames into the pre-trained detection model. The detection model detects the cyclist and bicycle in the video frame and selects them with a rectangular bounding box.

[0011] A size estimation module, which obtains the height and width of the cyclist and the bicycle based on the pixel coordinates of the rectangular bounding boxes of the cyclist and the bicycle in the detected video frame, and estimates the size of the cyclist and the bicycle;

[0012] A color estimation module estimates color information of a rectangular bounding box of a cyclist and a bicycle in a detected video frame based on a dominant color extraction algorithm, that is, the color of the cyclist and the bicycle;

[0013] A data storage module for storing the estimated size and color data of the rider and the bicycle;

[0014] The security module is triggered immediately when the rider is separated from the bicycle. When a rider reunites with the bicycle, the size and color of the rider who reunites with the bicycle are estimated based on the above steps, and the similarity of the size and color between the two riders is compared. If the similarity is higher than a preset threshold, the two riders are determined to be the same rider, and the module does not act. If the similarity is lower than the preset threshold, the two riders are determined to be different riders, and the size and color of the rider who reunites with the bicycle are saved, and an alarm is sounded at the same time.

[0015] Furthermore, the preprocessing module performs preprocessing of the video frames including scaling, color space conversion and denoising.

[0016] Furthermore, the detection model of the target detection module detects the cyclist and the bicycle in the video frame based on the open source Opencv DNN-AI basic language algorithm.

[0017] Furthermore, the security module compares the size similarity between the two riders based on the Euclidean distance, and compares the color similarity between the two riders based on the dominant color matching algorithm.

[0018] Furthermore, the security module saves the size and color of the rider who is re-attached to the bicycle and records the rider's violations, creates a violation record data structure including a timestamp, location information and violation type, and associates the rider's information with the violation record.

[0019] Furthermore, the security module alarms in the form of sound, light or notification, and the alarm time is ten minutes, which is timed by multi-threading or a timer.

[0020] Compared with the prior art, the present invention has the following beneficial effects:

[0021] The present invention collects video frames through a camera, detects the rider and the bicycle in the video frame through a target detection module, estimates the size and color of the rider and the bicycle in the video frame through a size estimation module and a color estimation module, and automatically triggers a security program when the rider is separated from the bicycle. The security module compares the similarity of size and color between the rider who reunites with the bicycle and the previous rider to determine whether they are the same rider, so as to determine whether to alarm or not. Compared with the prior art, the present invention can provide a new implementation plan for security, and at the same time, the security accuracy rate is high. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 This is a framework diagram of a security system for collecting distortion-free images for analysis based on a spherical camera according to the present invention;

[0023] In the figure: 1. Initialization module; 2. Acquisition module; 3. Preprocessing module; 4. Target detection module; 5. Size estimation module; 6. Color estimation module; 7. Data storage module; 8. Security module. DETAILED DESCRIPTION

[0024] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0025] See also Figure 1 The present invention provides the following technical solution: a security system based on a spherical camera collecting distortion-free images for analysis, comprising:

[0026] Initialization module 1, camera parameter initialization;

[0027] Acquisition module 2, the camera acquires video frames;

[0028] Preprocessing module 3, preprocessing the collected video frames;

[0029] The object detection module 4 inputs the preprocessed video frame into the pre-trained detection model, and the detection model detects the cyclist and the bicycle in the video frame and selects them with a rectangular bounding box;

[0030] A size estimation module 5 obtains the height and width of the cyclist and the bicycle based on the pixel coordinates of the rectangular bounding box of the cyclist and the bicycle detected in the video frame, and estimates the size of the cyclist and the bicycle;

[0031] A color estimation module 6 estimates color information of a rectangular bounding box of a cyclist and a bicycle in a detected video frame based on a dominant color extraction algorithm, that is, the color of the cyclist and the bicycle;

[0032] The specific steps of the dominant color extraction algorithm to estimate color information include:

[0033] The RGB color space of the rectangular bounding boxes of the cyclists and bicycles detected in the video frames is evenly divided at certain intervals to form a number of color cubes;

[0034] Use the center color of the cube as the mapping value for the color that falls within the cube;

[0035] Count the frequency of each color appearing in the rectangular bounding boxes of cyclists and bicycles in the detected video frames;

[0036] Select the color that appears most frequently as the color of the rectangular bounding box of the cyclist and bicycle in the detected video frame;

[0037] A data storage module 7 stores the estimated size and color data of the rider and the bicycle;

[0038] The security module 8 is triggered immediately when the rider is separated from the bicycle. When a rider rejoins the bicycle, the size and color of the rider who rejoins the bicycle are estimated based on the above steps, and the similarity of the size and color between the two riders is compared. If the similarity is higher than a preset threshold, the two riders are determined to be the same rider, and the module does not act. If the similarity is lower than the preset threshold, the two riders are determined to be different riders, and the size and color of the rider who rejoins the bicycle are saved, and an alarm is sounded at the same time.

[0039] Specifically, the preprocessing module 3 performs preprocessing on the video frame including scaling, color space conversion and denoising.

[0040] Specifically, the detection model of the target detection module 4 detects the cyclist and the bicycle in the video frame based on the open source Opencv DNN-AI basic language algorithm.

[0041] Specifically, the security module 8 compares the size similarity between the two riders based on the Euclidean distance, and compares the color similarity between the two riders based on the dominant color matching algorithm.

[0042] The specific steps of comparing size similarity using Euclidean distance include:

[0043] Construct the values ​​of two dimensions into a vector;

[0044] Use the Euclidean distance calculation formula to calculate the distance between two vectors. The expression is:

[0045]

[0046] Where: (x1, y1) represents the A coordinate, (x2, y2) represents the B coordinate;

[0047] The specific steps of the main color matching algorithm to compare color similarity include:

[0048] Convert the two colors to be compared from the RGB color space to the HSV color space and record them as Color1 (H1, S1, V1) and Color2 (H2, S2, V2) respectively;

[0049] Use the Euclidean distance calculation formula to calculate the distance between two colors in the HSV color space. The expression is:

[0050]

[0051] Wherein: ΔH represents the hue difference, which is obtained by calculating the difference between H1 and H2 and taking the absolute value; ΔS represents the saturation difference; ΔV represents the lightness difference, which is obtained by calculating the difference between S1 and S2 and between V1 and V2.

[0052] Specifically, the security module 8 saves the size and color of the rider who is re-attached to the bicycle and records the rider's violations, creates a violation record data structure including a timestamp, location information, and violation type, and associates the rider's information with the violation record.

[0053] Specifically, the security module 8 alarms in the form of sound, light or notification, and the alarm time is ten minutes, which is timed by multi-threading or a timer.

[0054] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A security system based on a spherical camera collecting distortion-free images for analysis, characterized in that: include: Initialization module (1), camera parameter initialization; Capture module (2), the camera captures video frames; A preprocessing module (3) performs preprocessing on the collected video frames; The object detection module (4) inputs the preprocessed video frame into the pre-trained detection model, and the detection model detects the cyclist and the bicycle in the video frame and selects them with a rectangular bounding box; A size estimation module (5) obtains the height and width of the cyclist and the bicycle based on the pixel coordinates of the rectangular bounding box of the cyclist and the bicycle detected in the video frame, and estimates the size of the cyclist and the bicycle; A color estimation module (6) estimates color information of a rectangular bounding box of a cyclist and a bicycle detected in a video frame based on a dominant color extraction algorithm, that is, the color of the cyclist and the bicycle; A data storage module (7) for storing the estimated size and color data of the rider and the bicycle; The security module (8) is triggered immediately when a rider is separated from the bicycle. When a rider rejoins the bicycle, the size and color of the rider who rejoins the bicycle are estimated based on the above steps, and the similarity of the size and color between the two riders is compared. If the similarity is higher than a preset threshold, the two riders are determined to be the same rider, and the module does not act. If the similarity is lower than the preset threshold, the two riders are determined to be different riders, and the size and color of the rider who rejoins the bicycle are saved, and an alarm is triggered at the same time.

2. According to claim 1, a security system based on a spherical camera collecting and analyzing distortion-free images is characterized in that: The preprocessing module (3) preprocesses the video frames including scaling, color space conversion and denoising.

3. The security system based on the spherical camera collecting and analyzing distortion-free images according to claim 1, characterized in that: The detection model of the target detection module (4) detects the cyclist and the bicycle in the video frame based on the open source Opencv DNN-AI basic language algorithm.

4. The security system based on the spherical camera collecting and analyzing distortion-free images according to claim 1, characterized in that: The security module (8) compares the size similarity between the two riders based on the Euclidean distance, and compares the color similarity between the two riders based on the dominant color matching algorithm.

5. The security system based on the spherical camera collecting and analyzing distortion-free images according to claim 1, characterized in that: The security module (8) saves the size and color of the rider who is re-attached to the bicycle and records the rider's violations, creates a violation record data structure including a timestamp, location information and violation type, and associates the rider's information with the violation record.

6. The security system based on the spherical camera collecting and analyzing distortion-free images according to claim 1, characterized in that: The security module (8) can generate an alarm in the form of sound, light or notification. The alarm time is ten minutes and is timed by a multi-thread or timer.

Images