Chung Cheng University Institutional Repository
Not a member yet
889 research outputs found
Sort by
軟體開發模式(SCRUM)實作異質系統整合與分析：以差假整合流程為例;Heterogeneous System Integration and Analysis Based on Scrum：Designing Absence Request Form Flow
No full text[[abstract]]本論文針對本校的公文系統及差勤系統實作整合及界接過程，提出使用在網路服務相關技術遠端程序呼叫（Remote Procedure Call，RPC）及服務導向架構（Service-oriented architecture，SOA）分析說明；並提出在整合兩個系統時使用的方案。於整個實作過程專案管控部份，也針對相關軟體工程開發模式上的研討，瞭解瀑布式及Scrum等系統開發模式，並以Scrum為基礎的軟體開發模式導入在公文系統端程式開發部份進行專案控管。藉由本論文提出的相關方案方法，透過一個實際的應用案例針對異質系統的整合方案、專案管理方法的改善及實作，既讓本校的差假電子作業上，顯著的更有效率。整合系統完成後，在假單申請量，每年處理量超過45,000件假單。而服務人次量每年也超過210,000人次以上；此外，也讓平均的假單處理時間在40小時左右可以完成
In this thesis, we introduce related analysis about Web Service between Remote Procedure Call and Service-oriented architecture in order to integrate Document On-line Submission and Approval System and Attendance Management System in National Chung Cheng University. Meanwhile, we propose the strategy to integrate these two different systems. By studying software engineering and development ways, we understand the advantages and disadvantages between Waterfall and SCRUM. Therefore, we adopt SCRUM to manage the development in Document On-line Submission and Approval System. Through related methods proposed by this thesis. We accomplish the integration of Heterogeneous System Integration. We have made the attendance management more efficient. After the integration, the number of absence request processed are over 45,000 every year. The system provides over 210,000 accesses every year, too. At the meantime, the average completed time of absence request flow is around 40 hours
用於放射治療之多視角即時呼吸導引系統;Multiple View-Angle Real-Time Breathing Guidance System for Radiotherapy
No full text[[abstract]]衛福部 103 年公布惡性腫瘤已連續33年被列為國人的主要死因，治療惡性腫瘤常見的方法是放射線治療，使用高速電子撞擊金屬靶產生出的高能量X光照射惡性腫瘤，阻止其異常的生長速度達到治療效果。在放射治療過程中光束應該要集中在腫瘤上，然而有放療相關文獻提到病患的呼吸運動會改變內臟組織和惡性腫瘤的相對位置，導致放射治療的效果不如預期，最壞的情況可能是惡性腫瘤照射的劑量不足，而且還傷害到正常組織器官。為了避免預期外的腫瘤位移情況發生，臨床常見方法是利用呼吸門控的特性，為病患的呼吸深度設定一段門檻值，當深度達到門檻值才啟動放射光束治療，基於此機制我們提出一個低成本且能提高放射線治療效果的多視角影像呼吸導引系統，硬體上使用了數個常見的影像擷取裝置和一台電腦，搭配一套使用多種影像處理技術的系統，即時追蹤病患在放療過程中因呼吸運動產生的胸腹部形變，並與當初利用電腦斷層做腫瘤定位時的呼吸深度快照記錄做比較。兩個時間點的胸腹部形變曲線會重疊顯示在病患配戴的虛擬實境眼鏡上，病人可利用這些影像資訊控制當下的呼吸深淺以匹配腫瘤定位時的記錄，使腫瘤位置盡量與當初制定治療計畫所用的參考影像相對齊，進而減少正常組織或器官受到不必要的照射。
According to the data of Ministry of Health and Welfare announced in 2014, malignant tumor has been the main cause of death for 33 consecutive years in Taiwan. A common treatment of malignant tumors is radiation therapy, which uses high-energy radiation beam to stop tumor growth and achieve therapeutic effect. The radiation beam should stay focusing on the malignancy during radiotherapy. However, previous studies showed that patients' respiratory movements alter the relative position between the visceral tissue and malignancy, resulting in less outcomes and extra damages.In order to avoid the occurrence of malignant tumor displacement, the common clinical solution is to set a threshold for the depth of patient's breath, and start the radiation beam illumination only when such depth reaches the threshold. Based on this mechanism we propose a “multi-perspective video-based breath guidance system”, which is low-cost and can improve the efficacy of radiation therapy. The hardware part includes several common imaging devices and a computer, along with a system software that includes a variety of image processing algorithms. This system can immediately track the thorax and abdomen deformations caused by respiratory movements in real time during radiotherapy, and compare with the original snapshot record of respiration depth when computer tomography was taken to locate the tumor. The thoracic and abdominal deformation curves at the two timing will be overlaid and projected on the patient's virtual reality headset. This information can be used to control the depth of the current breath to align with the one recorded during CT imaging, and reduce the damage of the normal tissue or organ from unnecessary irradiation
物聯網智能平台;A Smart Platform of IoT
No full text[[abstract]]在現代社會，物聯聯是一種趨勢，在物聯網蓬勃發展的同時，各家企業都有自己的感應產品；製造商也發展了自己品牌的感控設備，但是卻沒有一個標準的平台來整合這些感控設備所以我想開發一個整合平台來整合這些感控設備解決這個物聯網系統的問題。因此，本論文強烈地意識到目前的物聯網系統需要一個靈活，可擴展的物聯網平台來整合這些琳瑯滿目的感控設備。所以在這個平台中，我構建和發展了一個智能平台，平台可以動態添加各種感控設備，目前實體感控設備建設了，包括空調，燈光，亮度，風扇，溫度，濕度和虛擬感控設備有時間，股市，天氣，警戒，貨幣匯率等相關感測和控制設備，樹莓派和Arduino，分散的構建在環境中負責環境變量的檢測，無線網路則負責傳送平台的指令與接收偵測與判讀結果，平台則主要負責管理每個樹莓派和Arduino，分別表示其實際檢測和識識資料的傳輸。這個平台，可以作為物聯網規定的標準規格、操作方式及相關技術。提供相關行業或系統開發商在實際生產智能平台當一個值得參考的藍圖。
Under the vigorous development of the IoT, the manufacturers have developed their own brand of sensing equipment but do not have a standardized platform to integrate these devices. I would like to develop an integrated platform to level the integration of sensing devices to solve this IoT system. In modern society, the IoT is a trend, but companies have their own sensing products. Therefore, this thesis is strongly aware of the need for a flexible and scalable Internet networking platform to integrate these dazzling sensors.So in this system, the construction and development of a network of intelligent platform, the platform system to dynamically add all kinds of sensing and monitoring equipment, the current construction of the physical sensors, including air condition, light, brightness, fan, temperature, humidity, and virtual sensor time, stock market, weather, alert, currency exchange rate, and other related detection and control systems. Integrated wireless network, Raspberry Pi and Arduino, construct a monitoring system distributed in the sensing environment, Raspberry Pi and Arduino are responsible for detecting environmental variables, the wireless network is responsible for the transmission and reception control center detection and interpretation of the instructions, the control center is mainly responsible for the management of each raspberry Pi and Arduino, respectively, to indicate its actual detection and identification of data transmission. The design of a network of intelligent platform, as well as the Internet of each component in the registration of standard specifications and mode of operation and related technologies, as the provision of related industry or system developers in the actual production of intelligent platform when a worthy of reference blueprint
應用自動學習的深度相關性特徵於影像風格分類;Image Style Classification based on Learnt Deep Correlation Features
No full text[[abstract]]在電腦視覺研究領域中，大多數影像或影片的分類研究都專注於視覺特徵或是語意內容分析上，但其實有一些隱藏在影像中的特徵，像是影像中帶給人的情感或情緒的表達，對影像的分析也很有幫助。雖然人類能夠很容易的就感覺到這些特徵，卻很難給這些特徵一個明確的定義，或設計成一套演算法準確偵測出來。除了情感之外，影像的風格也是一種隱藏在影像中的特徵，目前在藝術史上已經被廣泛研究，但在電腦視覺中較少研究者注意，因此在本論文中我們將專注於影像風格分析的研究上。 本論文主要的目標是全面性地探討深度相關性特徵，並應用於影像風格分類系統上。首先，基於過去的研究發現計算特徵圖之間的相關性能有效的描述影像的紋理特徵，所以我們設計了許多不同的深度相關性特徵，將之應用於風格分類問題上。而我們所使用的特徵圖除了擷取自單一卷積層的特徵圖之外，也同時比較與使用不同卷積層特徵圖之間的差異。在初步的實驗中我們驗證了深度相關性特徵的有效性，此外我們進一步的透過神經網路建立一個學習架構，自動學習特徵圖之間的相關性。 在實驗的評測中，我們將深度相關性特徵廣泛地應用於多種影像風格資料集上。我們驗證所提出的特徵比過去常用的卷積神經網路特徵以及視覺特徵更能表達風格。最後，我們也將此特徵應用於畫家風格分類問題上，並再次驗證了其特徵的有效性。
Despite various studies on visual features and semantic concept detection, some image properties are difficult to extract, for the purposes of image/video classification or retrieval. Some bio-inspired properties, like sentiment and emotion, are apparently perceived by human, but are hard to be modeled in a computational way. In this work, we focus on image style property that emerges recently and is believed to be a promising extension of current classification/retrieval works. This paper presents a comprehensive study of deep correlation features on image style classification. Inspired by that correlation between feature maps can effectively describe image texture, we design and transform various such correlations into style vectors, and investigate classification performance brought by different variants. In addition to intra-layer correlation, we also propose inter-layer correlation and verify its benefit. After showing the effectiveness of deep correlation features, we further propose a learning framework to automatically learn correlations between feature maps. Through extensive experiments on image style classification and artist classification, we demonstrate that the proposed learnt deep correlation features outperform conventional CNN features and handcrafted deep correlation features by a large margin, and achieve the state-of-the-art performance
在內容核心網路中基於匈牙利演算法的內容放置機制;A Content Placement Scheme Based On Hungarian Algorithm For Content-Centric Network
No full text[[abstract]]當前的網路架構是在1960年代被發明的，而網路是讓兩個地方可以做溝通來使用的。而在過去幾十年間，網路的使用者有著巨量的成長。而網路的用途也從溝通的媒介變成讓使用者取得他們所想要的資料。即使目前有許多額外的設計來讓網路可以更符合使用者的需求，但是這些設計也使的網路架構變得更為複雜，更難以維護。而內容核心網路(Content Centric-Networks)是一個新穎的網路架構且符合使用者的需求。在內容核心網路中，內容核心網路給予每一個資料一個獨一無二的ID，而使用者可以透過這個ID直接取得他們所想要的資料。此外，在內容核心網路中，路由器可以擁有一定的儲存空間來暫時存放所經過的資料。藉由這項特色，路由器如果有存放使用者想取得的資料可以直接提供給使用者，如此一來可以有效的減少網路流量。然而，如何讓路由器有效的暫存資料是內容核心網路中一項很重要的議題。在這篇論文中，我們提出一個基於匈牙利演算法的內容放置機制。在我們所提出的方法中，我們考慮了請求以及資料所傳遞的距離、資料的受歡迎程度以及資料的大小來決定資料所放置的位置。除此之外，我們同時也掌控了資料在網路中所放置的份數，減少過多冗餘的資料。透過實驗結果，我們的機制比目前現有的機制有著較好的表現結果。
The network was create in the 1960s, and its goal is to contact two place. As time goes by, the users of network had increase very fast in the last decades. The goal of network also change to that provide data which is user want. Even though there are some patch to make network provide stable service, it make the network architecture more complicated and hard to maintain. Content Centric Network is a new paradigm of networking to meet the needs of users. It given every data a unique ID that make users can request the data directly. It also let routers has the cache capability to cache data. This feature make router can provide the data and also reduce the network traffic. However, how to cache data effectively is a big issue. In this paper, we model this cache problem as an assignment problem and propose a cache scheme which is based on the Hungarian Algorithm. Our cache scheme consider distance, content popularity and data size. We also allocate the number of data duplicates by considering their potential importance. We also make a simulation to verify. As shown in the simulation, our cache scheme has better performance than the others scheme
個人化電子商務搜尋引擎設計與實作;The Design and Implementation of Personal EC Search Engine
No full text[[abstract]]本論文實作一個個人化商品搜尋引擎。本搜尋引擎可以依據使用者對商品的興趣來調整搜尋結果的順序，以便使用者更方便地找到有興趣的商品。本系統包含一個商品資訊爬蟲，每天到台灣熱門的電子商務服務網站爬取商品資訊，還有一個Chrome瀏覽器外掛程式，藉由分析使用者在EC網站的瀏覽、搜尋行為來取得使用者對商品項目的興趣。有了使用者的興趣資訊，搜尋引擎在計算搜尋結果時，會參考使用者的喜好，將使用者較有興趣的商品項目排在較高的名次來輸出。本系統也提供商品追蹤功能，如果使用者追蹤一個商品項目，則當該商品項目有價格變動時，系統會主動通知使用者。
In this thesis, a personal product search engine is developed that can help the users find wanted product info more easily by a personalized search result ranking method. The system includes a crawler to crawl product information from popular EC sites in Taiwan daily, and we obtain the user's interest by a chrome extension that will track user's behavior in EC sites. Such information is then used to adjust the ranking of search result such that the product of more interest to the users are placed in higher positions. The system also provides a product tracking function. When a product is tracked by a user, the system will notify the user whenever such product has a price change
基於Arduino平台比較MQTT和WebSocket通訊協定表現;The comparison of MQTT and WebSocket performance, with the use of Arduino
No full text[[abstract]]物聯網的概念在業界並不陌生，最早在1999年就提出來了。物聯網的定義十分簡單，就是把所有物品通過射頻識別等信息傳感設備與互聯網連接起來，實現智能管理。物聯網的核心和基礎是互聯網，不過用戶端不僅局限於個人電腦，而是延伸到任何需要實時管理的物品和物品之間。如何有效的提升設備間傳輸的效率以及降低耗電成本來提高使用者使用智慧居家設備的意願，這是本文中希望可以達到的效果。本文主旨主要是使用三種網路通訊協定，包含一般HTTP通訊協定、MQTT (MQ Telemetry Transport or Message Queue Telemetry Transport)、以及WebSocket，比較其通訊模式的效率封包大小以及省電效果，並且透過實際在Arduino平台的操作，來得到確實認證的結果，主要會使用Arduino也是因為它可以接收各種感測器的輸入來源感測環境和其他的執行器來影響周圍環境，可以做出各式各樣的運用。也確實經過了電池的量測和封包的側錄，兩方都有明顯的差異性，證明了MQTT與WebSocket相較於一般HTTP協定的省電節能與通訊效率的提高，希望可以更使得物聯網的生活更貼近大家的生活，更為世界提供便利。
The concept of Internet of Things in the industry is no stranger. It first comes out at the 1999. The definition of Internet of Things is very simple, that is, all items are connected with the Internet through radio frequency identification and other information sensing equipment to achieve intelligent management. The core and foundation of the Internet of Things is the Internet, but the client is not limited to personal computers, however extends to anything that requires real-time management from items to items. How to improve the efficiency of transmission between devices and reduce the power consumption to improve the user's desire of using intelligent home equipment, which is expected in this paper.There are three internet protocols studied in this thesis, which are HTTP communicating protocol, MQTT (MQ Telemetry Transport or Message Queue Telemetry Transport) and WebSocket. We compare the efficiency of its communication mode packet size and power-saving effect, and through the actual operation in the Arduino platform, to get the exact results of the certification. Mainly arduino is used because it can sense the environment by receiving input from variety of sensors and can affect its surroundings by other actuators, can make a wide range of applications.Through the battery measurement and packet tracing and analysis, both sides have significant differences, that prove MQTT and WebSocket compared to the HTTP to save energy and improve the efficiency of communication. In order to make Internet of Things more close to everyone's life, it will provide convenience life to people
非同步工作週期無線感測網路具廣域延遲資訊路由協定;Extensively Delay-Aware Routing in Asynchronous Duty Cycle Wireless Sensor Networks
No full text[[abstract]]在無線感測網路中，能源效率之設計是一個重要的議題且已經有多年的研究。為了節省能源，工作週期的機制已被提出來解決能源問題。工作週期意思是感測器節點能透過它們的工作時間表來決定在短期間內維持醒著以進行資料傳送，相反的，感測器節點會利用剩餘期間進入睡眠以達到省電。工作週期又可以分成兩種，同步工作週期與非同步工作週期。在同步工作週期中，感測器節點會彼此協同在相同期間醒來以傳送資料，然而，為了達成這個目的，感測器節點必須交換彼此的工作時間表造成額外的封包成本。因此，提出了非同步工作週期。在這種型態中，感測器節點會獨立決定它們的工作時間表，所以容易佈建且無額外成本。但是，問題是感測器節點在有資料要傳時需要等待其中一個鄰居節點醒來以接收資料，在論文中我們稱這段等待時間為睡眠延遲。如果在每個節點間都有睡眠延遲，將會造成點到點延遲。有一先前研究就提出了分散式的方法來解決這個問題。他們利用sink節點透過生成樹廣播封包到整個網路讓每個感測器節點收集延遲資訊，所以每個感測器節點能得知如果它們透過鄰居節點傳送資料到sink節點的點到點延遲。然而，每個感測器節點如果透過生成樹來收集只能得到sink節點方向的延遲資訊。因此，我們在這篇論文中提出了廣域延遲資訊路由協定。首先，我們在感測器節點有相同工作週期的簡單環境提出了最佳解。再者，我們在感測器節點有不同工作週期的環境再提出兩個不同的路由協定。模擬實驗結果顯示出我們方法比先前研究有更短的點到點延遲。
In wireless sensor networks, energy efficient design is an important issue and has been researched in many years. In order to save energy, duty cycle scheme is proposed to solve the energy problem. Duty cycle means the sensor nodes can decide to keep awake for data transmission in a short time period according to their working schedule, on the other hand, for the remaining time period, sensor nodes turn into sleep mode for saving energy. It is divided into two types, synchronous duty cycle and asynchronous duty cycle. In synchronous duty cycle, sensor nodes cooperate with each other to stay awake in the same time period to transmit data, however, to achieve this purpose, sensor nodes need to exchange their working schedule so that there is additional packet overhead. Therefore, asynchronous duty cycle is proposed. In this type, sensor nodes determine their working schedule independently, so it is easy to deploy and there is no extra cost. However, the problem is that sensor nodes have to wait for one of their neighbor nodes wake up to receive when they have data to send, we call the waiting time as sleep latency in this paper. If there is sleep latency between each hop, it will cause long end-to-end delay. A previous work proposed a distributed solution to solve this problem. They use sink node to broadcast a packet through a spanning tree to the whole network to collect delay information for each sensor node, so every sensor nodes can know the end-to-end delay if they transmit data to sink node through their neighbor nodes. Nevertheless, each sensor nodes can only get the delay information of the direction to sink node if they collect the delay information through a spanning tree. Therefore, we propose Extensively Delay-Aware Routing in this paper. First, we propose an optimal solution in easy environment, which sensor nodes have the same working cycle. Second, we further propose two different routing protocols in different working cycle of sensor nodes environment. Simulation results show that our method has lower end-to-end delay than previous work
藉由置換空間重組縮短開機後反應時間的快速開機技術;Improving Response Time by Reordering Swap Sequence for Fast Booting
No full text[[abstract]]休眠式快速開機能大幅縮短裝置開機所需時間，而其中休眠檔的大小是影響系統回復時間的重要因素之一，休眠式快速開機使用Swap Before Hibernate機制，能使休眠檔大幅縮減進而縮短開機時間。然而在下次系統啟動時，因應用程式執行所需的分頁已被大量換出，導致系統啟動後的前幾秒內會有大量的缺頁錯誤，需要發出實體I/O去置換空間中取回所需的分頁，造成開機後幾秒內會有短暫的延遲現象。本論文提出了利用置換空間重組的方法，藉由記錄頁面換入的順序，將存取關聯性高的頁面叢集化，並於下次休眠前重新排序頁面寫出的順序，藉此優化休眠式快速開機喚醒後之反應時間，提升快速開機的使用者體驗
使用華氏碼並建構於可程式邏輯板之人體通道傳輸收發器;An FPGA-based Transceiver for Human Body Channel Communication using Walsh Codes
No full text[[abstract]]在人口老化及現代科技的發展之下，可攜式醫療電子設備及可隨身佩戴、隨時測量的電子裝置越來越多。這類型的裝置大多是使用無線的方式進行傳輸，像是常見的藍芽傳輸、ZigBee。但是這些技術有很高的耗電量以及因為很多裝置也使用相同頻帶造成嚴重的干擾的問題。因此，以人體通道作為傳輸媒介的技術被提出。人體通道傳輸可以避免干擾的問題、能提供較穩定的傳輸，同時其功率消耗比較低。因為以上的優點，我們提出一個以人體通道傳輸的傳收器。 Virtex-7 VC707 evaluation board (EVB) 板是一張具有穩定且高速工作頻率的 EVB板，我們把人體通道傳收器建構於其上並且以華氏碼作為調變碼。華氏碼具有正交特性，所以可以容忍低於一半華氏碼長度的錯誤位元資料能被正確解調。除此之外，我們選取長度為2的continuous identical digits (CID)華氏碼避免過長的CID出現造成接收端難以回復訊號。在bit error rate (BER)量測中，以140公分傳輸距離傳送108個位元資料時，在488 Kbps (3.12 Mcps), 976 Kbps (6.25 Mcps), 以及1.95 Mbps (12.5 Mcps) 上並沒有任何錯誤資料發生。在3.9 Mbps (12.5 Mcps) 的傳輸率下，BER達到10-5。 在本研究提出的人體通道傳輸收發器，基於傳輸穩定度的提高，我們也提供直接傳輸多媒體資料的模式，在電腦上把圖片分解經由universal asynchronous receiver/transmitter (UART) 介面送到傳送器，經由最長為140公分、1.95 Mbps的速度傳到人體通道後，能正確地把它接收回，最後可以重建回原來完整的圖片。
In recent years, the portable healthcare devices are more and more popular. The wireless transmissions are widely used in these devices, such as Bluetooth and ZigBee. However, these techniques have drawbacks of high power consumption and have interferences in the shared frequency band, Therefore, body channel communication (BCC) is proposed because it has relatively low power consumption and has less interferences. The proposed BCC transceiver which using Walsh codes as modulated codes is implemented on the Vurtex-7 VC707 evaluation board (EVB). The Walsh codes are orthogonal codes and the maximum number of error bits that can be tolerated can be up to half of the code’s length. Moreover, the CID of the Walsh codes are limited to 2 to avoid the no data transition situation which makes the receiver difficulty to recover the clock and data. There are no error bits occurred when transmitting 108 bits at 140 cm distance and at data rate of 488 Kbps (3.12 Mcps), 976 Kbps (6.25 Mcps), and 1.95 Mbps (12.5 Mcps). Moreover, the BER is 10-5 at 3.9 Mbps (12.5 Mcps). In the proposed BCC transceiver, because of high data reliability, the multi-media transmission is supported. The image can be completely transmitted and received through 140 cm distance and the maximum data rate is 1.95 Mbps