【摘要】：計算機(jī)視覺在現(xiàn)代信息化產(chǎn)業(yè)中得到了迅猛發(fā)展，基于圖像內(nèi)容分析的技術(shù)已成功應(yīng)用在交通車輛分析、工業(yè)產(chǎn)品檢測、農(nóng)業(yè)病蟲害防治、交互式廣告機(jī)、可視化對講機(jī)和醫(yī)療病癥檢測等各領(lǐng)域各場合。在醫(yī)療康復(fù)領(lǐng)域，目前主要以機(jī)器人輔助康復(fù)和人工輔助康復(fù)兩種方式。機(jī)器人輔助康復(fù)系統(tǒng)因其造價昂貴、維護(hù)成本高、智能性單一在實(shí)際的推廣應(yīng)用中市場慘淡；人工輔助康復(fù)是對患者康復(fù)最佳選擇，但現(xiàn)實(shí)中待康復(fù)的病患對康復(fù)教練的需求遠(yuǎn)大于供方。提供一種具有康復(fù)效果同時滿足康復(fù)病患購買能力和使用方便性的系統(tǒng)是康復(fù)訓(xùn)練領(lǐng)域的重要課題。 本文針對上述問題，以計算機(jī)視覺為解決問題的核心，通過Kinect設(shè)備采集病患者的空間深度圖，重建病患康復(fù)訓(xùn)練的動作流。系統(tǒng)設(shè)定兩種運(yùn)動角色:康復(fù)教練和康復(fù)患者，通過對其示范和訓(xùn)練的動作流的評估反饋給醫(yī)生病患康復(fù)訓(xùn)練的進(jìn)度報表。另外開發(fā)了系統(tǒng)的客戶端和WEB版管理端提供給康復(fù)訓(xùn)練過程中各角色基于Kinect骨骼采集技術(shù)的康復(fù)訓(xùn)練模型和流程。具體的主要內(nèi)容如下： 1．研究康復(fù)訓(xùn)練的方法，提出了基于Kinect骨骼采集技術(shù)的康復(fù)方法。該方法通過對計算機(jī)視覺中深度圖像運(yùn)動的分析，采集了人體在康復(fù)訓(xùn)練時運(yùn)動的動作流。通過對康復(fù)教練和康復(fù)病患兩種角色的運(yùn)動采集，建立康復(fù)訓(xùn)練的正負(fù)樣本。 2．研究康復(fù)訓(xùn)練的辦公流，提出了三角色互動康復(fù)辦公業(yè)務(wù)模式。該辦業(yè)務(wù)模型將康復(fù)訓(xùn)練過程中的三種角色：醫(yī)生、病患、康復(fù)教練融合到訓(xùn)練過程中。并針對其業(yè)務(wù)功能設(shè)計了基于角色權(quán)限的管理系統(tǒng)。 3．研究康復(fù)訓(xùn)練采集時正負(fù)動作流相似度評估問題，提出了融合DWT算法和RANSAC算法模型的動作流相似度評估算法，有效的解決了病患訓(xùn)練效果的評估問題，給醫(yī)生提供了置信度高的輸出結(jié)果。 4．研究康復(fù)訓(xùn)練系統(tǒng)中的三維建模和場景構(gòu)建問題，針對視覺康復(fù)系統(tǒng)，動作流可在三維場景中通過動畫人物重現(xiàn)，，創(chuàng)新性的實(shí)現(xiàn)通過多終端異地同場景的康復(fù)訓(xùn)練效果。
[Abstract]:Computer vision has been developed rapidly in the modern information industry. The technology based on image content analysis has been successfully applied in traffic vehicle analysis, industrial product detection, agricultural pest control, interactive advertising machine, etc. Visual walkie-talkie and medical symptoms detection in various fields and occasions. In the field of medical rehabilitation, robot-assisted rehabilitation and artificial-assisted rehabilitation are two main methods at present. Because of its high cost and high maintenance cost, robot assisted rehabilitation system has a poor market in practical popularization and application. Artificial assisted rehabilitation is the best choice for patients' rehabilitation, but in reality, the demand for rehabilitation coaches is much larger than that of the supply side for the patients to be rehabilitated. It is an important topic in the field of rehabilitation training to provide a system which can satisfy the purchase ability and convenience of the patients with rehabilitation at the same time. Aiming at the above problems, this paper takes computer vision as the core to solve the problem, and collects the spatial depth map of the patient through Kinect device, and reconstructs the motion flow of the patient's rehabilitation training. Two kinds of sports roles were set up in the system: rehabilitation coach and rehabilitation patient. Through the evaluation of their demonstration and training motion flow, the progress report of the rehabilitation training was given to the doctor. In addition, the client side of the system and the management end of the WEB version are developed to provide the rehabilitation training model and process for the characters in the rehabilitation training process based on the Kinect bone acquisition technology. The main contents are as follows: 1. To study the methods of rehabilitation training, and put forward a rehabilitation method based on Kinect bone acquisition technology. Based on the analysis of depth image motion in computer vision, the motion flow of human body during rehabilitation training is collected. The positive and negative samples of rehabilitation training were established by collecting the two roles of rehabilitation coach and rehabilitation patient. 2. To study the office flow of rehabilitation training, and put forward the three-role interactive rehabilitation office business model. The business model integrates the three roles of rehabilitation training: doctors, patients, and rehabilitation coaches into the training process. And the management system based on role privilege is designed for its business function. 3. The problem of evaluating the similarity degree of positive and negative motion flow in rehabilitation training collection is studied, and an algorithm for evaluating the similarity degree of motion flow based on DWT algorithm and RANSAC algorithm model is put forward, which effectively solves the evaluation problem of patient's training effect. It provides doctors with a high level of confidence in the output results. 4. The problem of three-dimensional modeling and scene construction in rehabilitation training system is studied. Aiming at the visual rehabilitation system, motion flow can be reproduced in three-dimensional scene by animating characters, and the effect of rehabilitation training through multi-terminal and same scene can be realized innovatively.
【學(xué)位授予單位】：電子科技大學(xué)
【學(xué)位級別】：碩士
【學(xué)位授予年份】：2013
【分類號】：TP274.2

【參考文獻(xiàn)】

相關(guān)期刊論文 前5條

1 劉建武,葉志前,陸金芳;虛擬現(xiàn)實(shí)在醫(yī)學(xué)中的應(yīng)用進(jìn)展[J];國外醫(yī)學(xué).生物醫(yī)學(xué)工程分冊;2000年06期

2 王英;高松;李筠;;電子生物反饋療法在腦卒中偏癱患者中的應(yīng)用[J];現(xiàn)代中西醫(yī)結(jié)合雜志;2008年15期

3 孫秀偉;閻麗;李彥鋒;;虛擬現(xiàn)實(shí)技術(shù)(VR)在醫(yī)療中的應(yīng)用展望[J];醫(yī)療保健器具;2007年05期

4 韓雯;常鈞;;現(xiàn)代康復(fù)工程的發(fā)展概述[J];醫(yī)療裝備;2007年04期

5 謝財忠;腦卒中康復(fù)的理論和方法[J];中國康復(fù)理論與實(shí)踐;2002年11期

本文編號：2468473