MINISTRY OF EDUCATION AND TRAINING HO CHI MINH CITY UNIVERSITY OF TECHNOLOGY AND EDUCATION GRADUATION PROJECT AUTOMATION AND CONTROL ENGINEERING APPLICATION IMAGE PROCESSING IN THE IDENTIFICATION OF PEOPLE FALLING LECTURER: PhD. VU VAN PHONG STUDENT: PHAM DUC THIEN TRAN DAI LOC SKL 010346 Ho Chi Minh City, February 2023 HO CHI MINH CITY UNIVERSITY OF TECHNOLOGY AND EDUCATION FACULTY FOR HIGH QUALITY TRAINING GRADUATION PROJECT APPLICATION IMAGE PROCESSING IN THE IDENTIFICATION OF PEOPLE FALLING PHAM DUC THIEN Student ID: 17151033 TRAN DAI LOC 17151018 Major : AUTOMATION AND CONTROL ENGINEERING Advisor : VU VAN PHONG, PhD. Ho Chi Minh City, January 2023 THE SOCIALIST REPUBLIC OF VIETNAM Independence – Freedom– Happiness -------- Ho Chi Minh City, January 7, 2023 GRADUATION PROJECT ASSIGNMENT Student name: _________________________ Student ID: ___________________ Student name: Student ID: ___________________ __________________________ Student name: Student ID: ___________________ __________________________ Major: Class: ________________________ ________________________________ Advisor: ____________________________ Phone number: _________________ Date of assignment: Date of submission: _____________ _____________________ 1. Initial materials provided by the advisor: ___________________________________ 3.
Content of the project: _________________________________________________ 4. Final product: ________________________________________________________ CHAIR OF THE PROGRAM ADVISOR (Sign with full name) (Sign with full name) THE SOCIALIST REPUBLIC OF VIETNAM Independence – Freedom– Happiness -------- Ho Chi Minh City, January 7, 2023 ADVISOR’S EVALUATION SHEET Student name:. Content of the project:. Approval for oral defense? (Approved or denied) .) Ho Chi Minh City, January 7, 2023 ADVISOR (Sign with full name) THE SOCIALIST REPUBLIC OF VIETNAM Independence – Freedom– Happiness -------- Ho Chi Minh City, January 7, 2023 PRE-DEFENSE EVALUATION SHEET Student name:.
Name of Reviewer:. Content and workload of the project. Approval for oral defense? (Approved or denied) .) Ho Chi Minh City, January 7, 2023 REVIEWER (Sign with full name) THE SOCIALIST REPUBLIC OF VIETNAM Independence – Freedom– Happiness -------- EVALUATION SHEET OF DEFENSE COMMITTEE MEMBER Student name:. Name of Defense Committee Member:.
Content and workload of the project .) Ho Chi Minh City, January 7, 2023 COMMITTEE MEMBER (Sign with full name) Acknowledgements To begin with, we would like to express our deepest gratitude and respect to Ph. Vu Van Phong, who gave us many important perspectives during the time we worked on our business project. We truly thank you for working together and directing all of us during the time we have learned and completed this venture. Although there are many difficulties and errors, we still try to overcome and appreciate Ph.
Vu Van Phong meets a lot and has accurate information so that our team can progress and complete extended time. We would also like to express our deep gratitude to the Board of Directors of Ho Chi Minh City University of Innovation and Training and the instructors in the force of Electrical - Electronic Design and Quality Training Human Resources. high, who taught and created all favorable conditions. conditions and environment help us in the process of studying and implementing the project.
Student Contents Chapter 1: INTRODUCTION. Jobs to do. 4 Chapter 2: LITERATURE REVIEW .1 PROCESS TO APPLY A GAUSS FILTER .2 Fourier Spectrum- Quick Refresh .5 Gaussian Filter and low pass filtering .1 When to use Gaussian blur .2 The aspects of a Gaussian blur filter .3 How Gaussian blur works in image filtering. 11 Chapter 3: HARDWARE DESIGN .2 C920s PRO HD WEBCAM .4 Function in code:.5 Image Progessing- Smoothing .11 Python HTTP Server.
40 Chapter 5: EXPERIMENT RESULTS/ FINDINGS AND ANALYSIS. 44 Chapter 6: CONCLUSION AND RECOMMENDATIONS. 55 List of Figures Figure 1.1Among geriatric and adult patients hospitalized for the treatment of trauma injuries, the etiology of trauma.1Two Dimensional gaussian function .2 Some types of Raspberry Pi [14] .2 Raspberry Pi 4 diagram [18] .3 C920 PRO HD WEBCAM [19] .1Human body position with respect to ellipse while falling [22] .2 Symbols used in our algorithm [22].1 resize 2x2 mask to 2x4 .2 resize 2x2 mask to 3x2 .1 Salt and pepper Noise .1 The three-step process of erosion .2 Image has been fully processed .3 Result of process .4 Steps are used to divide the operations .5 Image has been fully processed .6 Result of process .1 Specification of web server [23] .3 Pushbullet application interface .1 HTTP server communication [24] .2 Token used to send notification .3Images stored in raspberry .1 The front of the model .2 Fall detection model .3Model on sketching .1 When people stand .2 when people sit.3 When people fall .4 notification from mobile device .5 System detects falling. 47 List of Tables Table 1.1 Plan from w1 to w15 .1 Specification of Raspberry Pi .1Color space conversion codes .1 Results of fall detections .2 Performance of system.
49 Abstract This paper will present a system to monitor and analyze the human condition using continuous image subtraction between frames combined with a gaussian filter to identify human behavior in a specific environment, thereby making predictions about the sudden state change leading to human falls. This article uses posture estimation and abrupt position changes to achieve a recognition rate of over 90% of human falls. Key word: Fall detection, Pose estimation 0 Chapter 1: INTRODUCTION The first chapter will give an overview of the topic.1 Problem Statement Frequent or inadvertent falls in people can result in injury or death, particularly in the elderly. The percentage of older adults hospitalized after a fall is shown in the table below: Figure 1.1Among geriatric and adult patients hospitalized for the treatment of trauma injuries, the etiology of trauma.
[1] Recent studies or experiments on protecting people from the risk of falls are receiving great attention from everyone. These systems are created through the application of image processing technology, the system will identify the behavior of people in a specific area so that the person can recognize the possibility of a fall early and send it. Notify tracking devices such as smartphones to alert loved ones in order. take timely response measures to avoid danger to people and relatives.
Objective Design a system to monitor sudden falls in the elderly continuously. Design a tracking application to be installed on a family member's smartphone so that they can receive direct and earliest notifications when an incident occurs. 1 The system design can monitor in a variety of environmental conditions to ensure over 80% accuracy in fall diagnosis. Constraints Due to economic conditions, this study only stops at using 1 camera to monitor the elderly and only stops at monitoring in a fixed space of the camera.
In addition, the tracking is not effective in the environment is too dimly lit. The algorithm's shortcomings include the inability to compare frames relative to one another rather than to the initial frame. However, when compared to other comparable frames, there is a lot of noise. The outcomes would be far better if it were feasible to lower the noise using additional strategies.
The algorithm requires complete sight of the human body and cannot function in occlusion. In order to get around this restriction, it could be necessary to install an extra camera in an obstructed region. It compares the subsequent frames to the initial frame, which is another drawback. As it constantly compares to the initial frame, it could not show all of the contours if the first frame is not empty.
The algorithm was not tested in a scenario with numerous people, but according to our study, no warning would be necessary. Jobs to do Research theory Simulate system in laptop Select electronic components Design construction Build prototype Fix bugs and perfect the pro 2 1.1 Plan from w1 to w15 No Jobs to do People W1 W2 W3 W4 W5 W6 W7 W8 W9 W10 W11 W12 W13 W14 W15 Lộc, 1 Research theory Thiện 2 Simulate system in laptop Lộc 3 Select electronic components Thiện Lộc, 4 Design construction Thiện Lộc, 5 Build prototype Thiện Lộc, 6 Fix bugs and perfect the product Thiện 3 1. Estimated Expense Table 1.2 Cost estimated No List of things Cost 1.000 vnd Parts for raspberry 700.Tools Courses for program 400.000 vnd 4 Chapter 2: LITERATURE REVIEW After the overview in chapter 1, the chapter will give an overview of the world's research on the fall problem and the theory used in the article.1 Introduction Falls are a common phenomenon in children and the elderly due to an imbalance of the body or due to a decline in the function of some organs on the human body. Falls can range from mild to severe, especially in the elderly.
Especially in this group of people, the cause of sudden falls comes from many dangers from common old age diseases, such as calcium deficiency in the bones, which leads to sudden fractures, more dangerous than heart or brain problems.There are many studies on this phenomenon but the most prominent is the study of the epidemiology of falls in older people[2].Currently, there is a lot of research on the causes of falls in adults, typically due to complications of old age diseases[3].There are many methods used to identify people falling in use today, such as Children location monitoring system (CLMS), Vehicle tracking device (VTD), The student tracking system[4].This paper will present a method of human fall detection based on the pose estimation tool and the sudden change of position in a short period of time. As previously indicated, single sensor or sensor fusion techniques were used in fall detection studies. Individual wearable sensors, individual ocular sensors, individual ambient sensors, and data fusion through sensor networks are the four basic categories into which data collection techniques are commonly categorized. While some research combines visual and environmental sensors, this survey article treats them as two distinct categories since visual sensors are becoming a more common detection method with the introduction of depth cameras (RGBD), like the Kinect.
Important physiological changes in the human body brought on by falls can be used as a fall detection criteria. Accelerometers, gyroscopes, glucometers, pressure sensors, ECG (Electrocardiography), EEG (Electroencephalography), or EOG (Electromyography) can all be used to measure various aspects of the human body. [5] The commonly used method to monitor health for the elderly is often used that is to wear tracking devices on many locations on the body, but these have the disadvantage that these devices use batteries and they must be charged when the power runs out, so additional human assistance is required. Recently, we have had a strong development of camera technology and image processing to create devices that can monitor the abnormal fall behavior of the elderly with just a smartphone.
Elderly people will be continuously monitored through cameras arranged in the house and will notify relatives as soon as the person being monitored has sudden falls through connected electronic devices.2 System review With the development of computer vision within the last 10 years, progressive strategies based on images and recordings have become common in drop discovery systems. There are three basic approaches when it comes to image-based drop locations: using images from a single camera, 3D inspection against multiple cameras, and 3D research based on multiple cameras combined with individual sensors. By placing ovals around the question and examining the object's movement history over multiple continuous contours, the algorithm distinguishes changes in the ellipse's evolution and shows between past , during and after the fall to distinguish the accident. In addition, multi- frame parallelism methods are connected to extend computation speed and real-time response.3 GAUSSIAN FILTER A Gaussian Channel may be a moo pass channel utilized for reducing commotion (tall recurrence components) and obscuring districts of a picture.
The track is actualized as an Odd measured Symmetric Bit (Plunge form of a Lattice) passed through each pixel of the Locale of Intrigued to urge the required impact. The part isn't tricky towards drastic color changes (edges) due to the pixels towards the center of the part having more weightage towards the ultimate esteem at that point the outskirts.