基於人體骨架特徵點座標配合物理方法推估失衡部位問題之研究

NCUIR > Engineering College > Graduate Institute of opto-Mechatronics > Electronic Thesis & Dissertation > Item 987654321/92315

Please use this identifier to cite or link to this item: http://ir.lib.ncu.edu.tw/handle/987654321/92315

Title:	基於人體骨架特徵點座標配合物理方法推估失衡部位問題之研究
Authors:	何承諭;He, Cheng-Yu
Contributors:	光機電工程研究所
Keywords:	機器視覺;特徵點座標;肌肉失衡;人體姿態估測;Machine Vision;Feature Point Coordinates;Muscle Imbalance;Human Body Posture Estimation
Date:	2023-07-25
Issue Date:	2024-09-19 15:46:10 (UTC+8)
Publisher:	國立中央大學
Abstract:	現在智慧型手機的普及很多人在使用時姿勢不夠正確的情況下，容易造成身體不平衡與肌肉失衡的現象產生，而且往往一般人都不知道自己哪裡有問題，直到痠痛問題發生才開始警覺，所以了解自己身體的姿態是有其必要性。而且目前檢測基本上利用X光進行，不僅成本高且會有傷身體的疑慮。所以希望利用人體姿勢估測的開源程式來推估人體姿態對肌肉失衡的影響，進而藉由提取人體骨架關鍵點計算傾斜角度，接著配合熱像儀及硬度計的物理方式進行輔助檢測確認相關資訊的可靠度。本研究分為幾個流程講解，第一必須先要了解一個人的姿態進而建立一套姿態評估系統利用MediaPipe提供的姿態估計模型，通過受測者拍攝正面、背面、左側、右側四個面向照片中提取關鍵點位置，使我們能夠準確地計算施測部位的傾斜角度。第二利用熱像儀及硬度計做物理輔助測量了解，分析施測部位的熱度分布觀察溫差之間造成的影響，而硬度計針對自然站及挺直站兩種姿態，選取重要的肌肉群進行量測。第三經過以上檢測方法配合受測者填寫的問卷了解肌肉失衡部位，綜合檢測結果與問卷內容交叉比對。參與實驗的案例中，受測者年齡介於22~27歲之間在自然站姿拍攝下經實驗結果統計，正面及背面計算出傾斜角度後可以看出當受測者慣用手為右手時，80%的受測者肩膀有左高右低的現象產生，然後側面統計結果觀察到接近100% 的受測者頸部都有上交叉症候群的現象，而骨盆方面大約有75% 的受測者有下交叉症候群等跡象。以上傾斜部位經過姿勢調整後再次檢測接近100% 的受測者都會有所改善。而在溫度檢測時觀察出年輕人兩側肩膀溫度相差不大。接著在硬度檢測方面可以看出姿勢調整後，原本肌肉緊繃的地方會有所放鬆的趨勢。所以藉由本系統日常即可透過手機拍攝在不限時間或是地點就可得知身體出現的偏差角度，提供受測者了解姿態方面的問題。 ;Nowadays, the popularity of smart phones is not enough to use the correct posture of many people, easy to cause the body imbalance and muscle imbalance phenomenon, and often people do not know where they have problems, until the pain problem occurs before they start to alert, so it is necessary to understand their body posture. The use of X-rays is basically used for testing, which is not only costly, but also can be harmful to the body. Therefore, we hope to use the open source program of human skeleton estimation to estimate the effect of human posture on muscle imbalance, and then calculate the inclination angle by extracting the key points of human skeleton, and then confirm the reliability of relevant information by physical methods of thermal imaging and hardness measurement. This study is divided into several processes. First, we must first understand a person′s posture to establish a posture evaluation system using the posture estimation model provided by MediaPipe, and extract the key points from four facing photos of the subject: front, back, left, and right, so that we can accurately calculate the tilt angle of the measured part. Secondly, we use thermal imaging camera and hardness tester to do physical assisted measurement to understand and analyze the heat distribution of the test area to observe the effect of temperature difference, while the hardness tester selects the important muscle groups for measurement in two postures: natural standing and upright standing. Thirdly, the above testing methods were combined with the questionnaires filled out by the subjects to understand the muscle imbalance areas, and the comprehensive testing results were cross-referenced with the questionnaire contents. In the case of the experiment, the subjects aged between 22 and 27 years old were photographed in a natural standing posture and the experimental results showed that 80% of the subjects had left high and right low shoulders when the subject′s preferred hand was the right hand. About 75% of the subjects showed signs of lower cross syndrome. Nearly 100% of the subjects showed improvement in the above tilted areas after postural adjustment and retesting. In the temperature test, it was observed that the difference in temperature between the two shoulders of young people was not significant and should be noted in daily life. There is a clear trend of relaxation in the area of muscle tension after posture adjustment for hardness testing. With this system, you can take a picture of your body at any time or place by cell phone to know the deviation angle of your body, so that you can understand the posture problem.
Appears in Collections:	[Graduate Institute of opto-Mechatronics] Electronic Thesis & Dissertation

Files in This Item:

File	Description	Size	Format
index.html		0Kb	HTML	26	View/Open

社群 sharing

Loading...