WO2018161849A1 - Alarm system for falling in water based on image water texture and method therefor - Google Patents

Abstract

Disclosed are an alarm system for falling in water based on an image water texture and a method therefor, comprising a video camera, an image pre-processing module, a feature extraction image module, a person recognition, positioning and tracking module, a fall in water water texture recognition, positioning and tracking module, a logic determination module and an alarm module, characterised in that video information photographed by the video camera is firstly processed by the image pre-processing module, then performing motion detection and feature extraction on the image pre-processed information through the feature extraction image module. The present invention uses a fall in water water texture as a key judgement feature for a person falling in water, such that recognition is faster and more accurate. All that is needed is that when the system recognises a fall in water water texture and a person at the same time, and the distance between the two objects is less than a threshold value, an alarm is triggered, reducing the probability of a false alarm from the system.

Description

Water drop alarm system based on image water pattern and method thereof Technical field

The invention relates to the field of intelligent alarms, in particular to a water falling alarm system based on image water ripple and a method thereof.

Background technique

At present, for the water drop alarm, the main method is to wear the sensor on the person. When the human body falls into the water, the sensor can detect the falling water information and alarm. Disadvantages: This method can only protect the person wearing the sensor, is not universal, and can not protect the life safety of ordinary people. Especially in some hydrophilic coastal areas or scenic spots.

Summary of the invention

The object of the present invention is to overcome the deficiencies of the prior art, and an object of the present invention is to provide a water-falling alarm system and method thereof based on image watermark, and to solve the problem that some hydrophilic coastal areas or scenic spots can be monitored in time without field personnel monitoring. The problem of people falling into the water.

The invention is achieved by the following technical solutions:

A water falling alarm system and method thereof based on image water pattern, comprising a video camera, an image preprocessing module, an extracted feature image module, a human body identification positioning tracking module, a water mark recognition positioning tracking module, a logic judgment module and an alarm module, The video information captured by the video camera is first processed by the image preprocessing module, and the image preprocessed information is subjected to motion detection and feature extraction by extracting the feature image module;

The information obtained by extracting the feature image module is simultaneously monitored by the human body recognition and location tracking module and the water mark recognition tracking and tracking module; the information obtained by the human body recognition location tracking and the water mark recognition tracking module is transmitted to the logic judgment. The module part performs the identification judgment. If the human body and the falling water pattern appear at the same time in the video, and the distance between the two is less than a preset threshold, it is judged that a person falls into the water within the coverage of the video, and an alarm is issued through the alarm module.

Further, the image pre-processing module includes an operation of denoising, enhancing, and enhancing the image.

Further, the extracted feature image module extracts pixel information whose gray value changes in the video information exceeds a threshold value, removes still pixels in the video, and reduces unnecessary data amount processing.

Further, the human body feature used by the human body recognition and location tracking module is a head feature, and the information obtained by extracting the feature image module is compared with the human head sample library to identify whether it is a human body.

Further, the water and water pattern recognition and tracking module adopts the water drop pattern of the human body as a basis for judging, and compares the information obtained by extracting the feature image module with the human body water and water sample database to identify whether the water is watery.

The invention is achieved by another technical solution as follows:

A watermarking method based on image watermark, comprising the steps of:

Step 1) Establish a human head sample library and a human body water and water sample library;

Step 2) The video information captured by the video camera is first processed by the image preprocessing module, and the image preprocessed information is subjected to motion detection and feature extraction by extracting the feature image module;

Step 3) The information obtained by extracting the feature image module is simultaneously monitored by the human body recognition and location tracking module and the water mark recognition tracking and tracking module;

Step 4) The information obtained by the human body identification and tracking and the water mark recognition tracking and tracking module is transmitted to the logic determination module for identification and determination;

Step 5) When both the human body and the falling water pattern appear in the video, and the distance between the two is less than a preset threshold, it is judged that a person falls into the water within the coverage of the video, and an alarm is issued through the alarm module, otherwise no alarm is issued.

Further, the image preprocessing module in the step 1) includes an operation of denoising, enhancing, and highlighting the image.

Further, the extracted feature image module in the step 1) extracts pixel information whose gray value changes in the video information exceeds a threshold value, removes still pixels in the video, and reduces unnecessary data amount processing.

Further, the human body feature adopted by the human body recognition and location tracking module in the step 4) is a head feature, and the information obtained by extracting the feature image module is compared with the human head sample library to identify whether it is a human body.

Further, in the step 4), the water and water pattern recognition and tracking module adopts the water drop pattern of the human body as a basis for judging, and compares the information obtained by extracting the feature image module with the human body water pattern sample library to identify whether the body falls into the water. Pattern.

Compared with the prior art, the present invention has the following advantages and beneficial effects:

The invention adopts the falling water pattern as the key discriminating feature of the human body falling water, and the identification is faster and more accurate. Only when the system recognizes the water mark and the human body at the same time, and the distance between the two objects is less than the threshold, the alarm is triggered, which reduces the probability of system false positives.

DRAWINGS

The drawings are intended to provide a further understanding of the embodiments of the present invention, and are not intended to limit the embodiments of the invention. In the drawing:

1 is a schematic diagram of the principle of an image-based water drop alarm system according to the present invention;

Marked and corresponding part names in the drawing:

1-Video camera, 2-image pre-processing module, 3-extract feature image module, 4-human body recognition and tracking module, 5-fall water pattern recognition and tracking module, 6-logic judgment module.

detailed description

The present invention will be further described in detail below with reference to the embodiments and the accompanying drawings. As a limitation of the invention.

Example 1

As shown in FIG. 1 , an image-based water drop alarm system of the present invention comprises: a video camera 1 , in a water area to be monitored, according to a water area, a plurality of video cameras are arranged in a range of 5 meters from the shore; No dead angle coverage; image preprocessing module 2; extraction feature image module 3; human body recognition location tracking module 4; water and water pattern recognition location tracking module 5; logic determination module 6; The video information captured by the video camera 1 first passes through the image preprocessing module 2, and the image preprocessing module 2 includes operations such as denoising, enhancing, and highlighting the image, so that the image information can better reflect the real situation and is easier to handle. Reduce the amount of calculations. The image pre-processed information is subjected to motion detection, and the feature image module 3 is extracted, and the pixel information in which the gray value changes in the video information exceeds the threshold value is extracted, the still pixels in the video are removed, and unnecessary data amount processing is reduced. The information obtained by the feature image module 3 is simultaneously subjected to the human body body position tracking module 4 and the water mark recognition position tracking module 5, wherein the human body feature used by the body recognition position tracking module 4 is a head feature, and the feature image module 3 is extracted. The obtained information is compared with the human head sample library to identify the human body; the human head sample library is a database established according to the characteristics of the human head, and an image is recorded on each orientation of the human head, and a database for extracting the feature of the part is extracted. The head picture taken by the video camera 1 is compared with the human head sample library to identify whether it is a human body.

The water mark recognition and tracking module 5 uses the water drop pattern of the human body as a basis for judging, and compares the information obtained by the motion detection and extraction feature image module 3 with the human body water and water sample bank to identify the water drop pattern of the human body. The human body water and water sample library is a database based on the water pattern formed by the human body falling water. Due to the special shape of the human body, after the person falls into the water, the water pattern formed by the stress reaction has a special shape, and the database is established by the human body. The information obtained by the identification and tracking module 4 and the watermark recognition positioning and tracking module 5 is transmitted to the logic determination module 6. If the human body and the watermark appear simultaneously in the video, and the distance between the two is less than a preset threshold, When the threshold is set to 0-1m, it is judged that there is a person falling into the water within the coverage of the video and an alarm is issued.

The invention adopts the falling water pattern as the key discriminating feature of the human body falling water, and the identification is faster and more accurate. Only when the system recognizes the water mark and the human body at the same time, and the distance between the two objects is less than the threshold, the alarm is triggered, which reduces the probability of system false positives.

Example 2

A watermarking method based on image watermark, comprising the steps of:

Step 1) Establish a human head sample library and a human body water and water sample library; the human head sample library is a database established according to the characteristics of the human head, image recording of various orientations of the human head, and extracting a database of feature extraction of the part. The head picture taken by the video camera 1 is compared with the human head sample library to identify whether it is a human body. The human body water and water sample library is a database established according to the water pattern formed by the human body falling water. Due to the special shape of the human body, after the person falls into the water, the water pattern formed by the stress reaction has a special shape, and the database is established.

Step 2) The video information captured by the video camera 1 is first processed by the image preprocessing module 2, and the image preprocessed information is subjected to motion detection and feature extraction by extracting the feature image module 3; the image preprocessing module 2 includes denoising the image. , enhancements, and enhancements.

Step 3) The information obtained by extracting the feature image module 3 is simultaneously monitored by the human body recognition location tracking module 4 and the watermark recognition tracking and tracking module 5; the feature image module 3 is extracted to change the gray value of the video information beyond a threshold. The pixel information is extracted, and the still pixels in the video are removed; unnecessary data processing is reduced.

Step 4) The information obtained by the human body recognition and location tracking module 4 and the watermark recognition tracking and tracking module 5 is transmitted to the logic determination module 6 for identification determination; and the step 4) is used by the human body identification and location tracking module 4 The human body feature is a head feature, and the information obtained by extracting the feature image module 3 is compared with the human head sample library to identify whether it is a human body. The water and water pattern recognition and tracking module 5 uses the water drop pattern of the human body as a basis for judging, and compares the information obtained by the extracted feature image module 3 with the human body water and water sample library to identify whether the water is watery.

Step 5) When both the human body and the falling water pattern appear in the video, and the distance between the two is less than a preset threshold, the threshold is set to 0-1m. Within this threshold, the watermark of the specification is likely to be formed by the human body falling water. Otherwise, it is not relevant, it is judged that there is a person falling into the water within the coverage of the video, and the alarm is issued through the alarm module 7, otherwise the alarm is not performed.

The invention adopts the falling water pattern as the key discriminating feature of the human body falling water, and the identification is faster and more accurate. Only when the system recognizes the water mark and the human body at the same time, and the distance between the two objects is less than the threshold, the alarm is triggered, which reduces the probability of system false positives.

The specific embodiments of the present invention have been described in detail with reference to the preferred embodiments of the present invention. All modifications, equivalent substitutions, improvements, etc., made within the spirit and scope of the invention are intended to be included within the scope of the invention.

Claims (10)

A falling water alarm system based on image water ripple, comprising a video camera (1), an image preprocessing module (2), an extracted feature image module (3), a human body recognition location tracking module (4), a water mark recognition tracking and tracking module (5) The logic judgment module (6) and the alarm module (7) are characterized in that the video information captured by the video camera (1) is first processed by the image preprocessing module (2), and the image preprocessed information Motion detection and feature extraction are performed by extracting the feature image module (3); The information obtained by the extracted feature image module (3) is simultaneously monitored by the human body recognition and location tracking module (4) and the water mark recognition tracking and tracking module (5); the human body recognition location tracking module (4) and the falling water The information obtained by the pattern recognition and tracking module (5) is transmitted to the logic judgment module (6) for identification and determination. If the human body and the water ripple appear simultaneously in the video, and the distance between the two is less than a preset threshold, It is judged that there is a person falling into the water within the coverage of the video, and an alarm is issued through the alarm module (7). The image watermark based water fall alarm system according to claim 1, wherein the image preprocessing module (2) comprises an operation of denoising, enhancing and enhancing the image. The watermark-based water fall alarm system according to claim 1, wherein the extracted feature image module (3) extracts pixel information whose gray value changes in the video information exceeds a threshold value, and removes the video. Still pixels. The watermark-based water fall alarm system according to claim 1, wherein the human body feature used by the human body recognition and tracking module (4) is a head feature, and the feature image module is extracted (3) The obtained information is compared with the human head sample library to identify whether it is a human body. The water-based water-based falling water alarm system according to claim 1, wherein the water-falling watermark recognition and tracking module (5) adopts a human body water-water pattern as a judgment basis, and the feature image module is extracted (3). The information obtained is compared with the human body water and water sample library to identify whether the body is watery. A method for water drop alarm based on image watermark, characterized in that the method comprises the steps of: Step 1) Establish a human head sample library and a human body water and water sample library; Step 2) The video information captured by the video camera (1) is first processed by the image preprocessing module (2), and the image preprocessed information is subjected to motion detection and feature extraction by extracting the feature image module (3); Step 3) The information obtained by extracting the feature image module (3) is simultaneously monitored by the human body recognition location tracking module (4) and the water mark recognition tracking and tracking module (5); Step 4) The information obtained by the human body identification and location tracking module (4) and the water mark recognition tracking and tracking module (5) is transmitted to the logic determination module (6) for identification and determination; Step 5) When both the human body and the falling water pattern appear in the video, and the distance between the two is less than a preset threshold, it is judged that the person falling into the water within the coverage of the video, and the alarm is performed by the alarm module (7), otherwise it is not performed. Call the police. The image watermark based water fall alarm system according to claim 6, wherein the image preprocessing module (2) in the step 1) comprises an operation of denoising, enhancing and enhancing the image. An image watermark based water fall alarm system according to claim 6, wherein the feature image module (3) in the step 1) extracts pixel information in which the gray value of the video information changes beyond a threshold value. , remove the still pixels in the video. The watermark-based water fall alarm system according to claim 6, wherein the human body feature used in the human body identification and tracking module (4) in the step 4) is a head feature, and the feature image is extracted. The information obtained by the module (3) is compared with the human head sample library to identify whether it is a human body. The water-based water-falling alarm system according to claim 6, wherein: the step 4) water-falling watermark recognition and tracking module (5) adopts the human body water-water pattern as a judgment basis, and the extraction feature is extracted. The information obtained by the image module (3) is compared with the human body water and water sample library to identify whether the human body is watery.

Images