Chung Cheng University Institutional Repository
Not a member yet
889 research outputs found
Sort by
以沙氏法案為基礎的企業資訊系統一般控制要項建構與實證
Full text link[[abstract]]隨著時代不斷的進步，資訊科技的發展也是日新月異，為了提升企業本身的競爭能力，許多企業都紛紛開始採用各種資訊系統，例如企業資源規劃系統、客戶關係管理系統、供應鏈管理統等。再使用這些資訊系統的時候，雖然為企業帶來相當大的便利，但是另一方面也因為資訊系統本身的複雜性使然，在安全性上也留下了一大隱憂。資訊系統的內部控制成為了一個相當重要的課題，而其中又以一般控制是最為必須所受到重視的，如果有良好的一般控制，勢必能夠提升資訊系統本身的安全性。
本研究第一階段透過相關文獻的探討及整理，採用紮根理論研究法找尋出資訊系統應有之內部控制作項目的相關變數，以證期局所訂定之「公開發行公司建立內部控制制度處理準則」中的控制作業規範建構出本研究之雛形。再經由擁有企業資訊系統維護、管理及稽核實務經驗專家們的修正與確認，試圖提出『企業資訊系統之一般控制要點』，以提供稽核人員對企業資訊系統進行資料品質稽核的作業依據。第二階段將以此理論端建立之雛型為基礎，以個案研究法驗證，確認本研究提出之控制項目的可用性及有效性。
藉由本研究所建構出十一個個構面共48項控制項目的成果，期望能夠提供一個參考準則，以協助企業建立起資訊系統上內部控制的完善管理機制，並提供稽核人員在進行企業資訊系統上的稽核時可以針對企業資訊系統應控管的內部控制要點進行有效查核，讓企業能夠降低在資訊系統使用上的風險。而在學術研究方面，也可作為後續相關研究的一個範例，讓有興趣對資訊系統進行研究者，可以針對不同的資訊系統去深入探討出彼此間的內部控制有何差異存在，以建構出專屬於特定資訊系統的內部控制項目
DsPIC Based Control System with CAN BUS for Robot Manipulators
Full text link本篇論文的研究主要是利用微控制器搭配DE2-70 嵌入式開發板為開發平台，實作出一套模組化的機器手臂控制系統。在論文中所使用的硬體發展平台是使用 Altera 公司的DE2-70 開發板搭配Microchip 所生產的DsPIC33FJ128MC710 為機器手臂的系統控制器。
在實作方面，整體硬體系統架構可以細分為主機端、NIOS II、DsPIC 微控制器三部分；主機端提供一個GUI 介面方便使用者進行逆向運動學座標輸入，並透過主機端UART 介面將座標傳送至DE2-70 開發板的NIOS II 系統中進行逆向運動學的運算，NIOS II 系統將逆向運動學各角度計算完畢後再將命令傳送至各軸的微控制器進行馬達的位置控制。
本文在DE2-70 開發板與DsPIC 微控制器之間的訊號傳輸的通訊介面是使用 CAN BUS 訊號傳輸匯流排技術，進行倆模組化系統之間的訊號傳輸，由於CAN BUS 技術傳輸訊號穩定、快速，相當適合用於大多數機器人的研究與開發。[[abstract]]This thesis exploits microcontrollers with the DE2-70 embedded development board for development platform to make a modularity robotic manipulator control system. The hardware development platform which is mentioned in the dissertation is DE2-70 development board from ALTERA Company with DsPIC33FJ128MC710 produced by microchip system controller for the robot manipulator.
In the implementation, the overall hardware system structure can be separated into three parts, GUI, NIOS II core process, and DsPIC microcontroller. The host provides a GUI for users to input inverse kinematics coordinates, and through the host UART interface, we can send the coordinates to the NIOS II system of DE2-70 development board to compute inverse kinematics. After calculating the axis angles of inverse kinematics, NIOS II system will send the instruction to the microcontroller to control the position of the motor.
In this thesis, the technology of CAN BUS signal transmission bus is used to transmit the signal between DE2-70 embedded development board and DsPIC microcontroller. Because of the CAN BUS technology with stability and fast quality, it is very suitable for the most research and development of robot
修改像素預測誤差得到高藏量的自適應式訊息隱藏方法;Adaptive Information Hiding with High Embedding Capacity Using Modification of Pixel Prediction Errors
No full text[[abstract]]隱寫術被廣分的使用在嵌入大量祕密資訊並同時保持著令人滿意的影像品質，其中最著名的技術便是預測誤差值的應用在隱寫術上。在這篇論文中，分別在可逆以及不可逆的隱藏中提出了兩種不同的方法來達到更高藏量且仍能有好的影像品質。首先，在不可逆式資訊隱藏，最低顯著位元藏入法結合像素誤差值藏入法中融入我們的影子像素預測法。第二則是在可逆式資訊隱藏，周圍像素預測誤差直方圖位移法中改良使用了混合預測誤差法以及二維直方圖成對像素誤差預測藏入法，在經過這兩種的改進後，就能得到更好的藏量與影像品質。第一個部分的改進，因為像素誤差值藏入法中的誤差越大能藏的數量越多的關係，我們利用影子像素預測法，選擇較大的誤差值，藉此增加了藏量，同時利用了奇偶性位元的關係進一步擴大藏量，因此得到了極高的藏量也擁有好的影像品質，第二部分，因為周圍像素鄰近的關係，我們利用了混合預測誤差法，使我們的預測更加精準，也就能夠獲得更多的峰值用來進行藏入，原本我們直接使用了雙峰值的直方圖誤差預測藏入法，得到了極高的藏量，但經過權衡我們使用了二維直方圖成對誤差預測藏入法，不但提升了藏量也擁有較好的影像品質。
Steganography is widely used to embed large amounts of secret information while maintaining well image quality. The most famous steganography technique is prediction error. In this paper, two different approaches have been proposed for reversible and irreversible data hiding to achieve higher levels of capacity and still have good image quality. First of all, in the irreversible data hiding, the least significant bit method incorporates the pixel value difference method into our shadow pixel prediction method. The second improve is the use of the mixed prediction error method and the surrounding pixel prediction error histogram shifting will be displaced by two-dimensional histogram pixel error prediction hiding method in the reversible information hiding. After these two improvements we can get better capacity and good image quality.The first part of the improvement, due to the larger the difference in the pixel value the greater the number of hidden capacity, we use the shadow pixel prediction method to select a larger difference value, thereby increasing the amount of peak point, while using the relationship of parity bits to further expand the capacity, we have obtained a high storage and also have good image quality. In the second part, du to the proximity of the value of neighboring pixels, we have used the mix prediction error expansion method to make our predictions more accurate, and we can have more peak point for hiding. We originally used the double peak-point histogram prediction error expansion method to get a very high level capacity, but after thinking, we used two-dimensional pairwise prediction error histogram shifting method has not only improved the capacity but also has good image quality.Keyword
以深度網路區域提案分類及回歸偵測漫畫中的文字區塊;Text Detection in Manga by Deep Region Proposal, Classification, and Regression
No full text[[abstract]]儘管在自然影像中的場景文字偵測(scene text detection)在學術研究中已經發展長遠，現存的自然影像文字偵測方法卻不適合直接應用在漫畫的文字偵測中，原因是漫畫中的文字有相當大的變異性與不同的上下文資訊。於本論文中，我們提出一個基於深層網路的漫畫文字偵測方法。此方法的主要偵測流程包含區域提案(region proposal)，特徵提取(feature extraction)，分類及回歸(classification/regression)，我們將它們整合在單一的網路中。我們也結合一個空間轉換(spatial transformer)的網路以改善偵測的準確度。這個網路最主要的目的是藉由空間上扭曲轉變特徵，以達成能夠讓網路中的分類更加準確。空間轉換網路的整合讓我們的偵測方法能學習漫畫文字空間轉換上的特徵關係，將特徵加以轉換，最終使分類網路能夠正確分類困難或是變形劇烈的文字。最後經由實驗證明，我們的偵測方法表現優於現存的作法。因此在漫畫的文字偵測上，本論文提供一個目前為止最先進的方法。
Though scene text detection for natural images has been studied for years, text in manga presents high variations and different contextual information, and existing scene text detection methods are not directly applicable. In this thesis, we propose an approach based on a deep network to detect text in manga. In this approach, primary processes of text detection, including region proposal, feature extraction, and classification/regression, are taken together in a single deep network. We also improve our approach by integrating a spatial transformer network. This network is proposed to deform the features spatially to make classification more accurate. In our work, the spatial transformer network is used to learn spatially deforming feature maps, and advances performance of the detector.The evaluation results show that our approach yields a big performance leap over the current state of the art, making it the leading method in manga text detection
中醫藥材知識本體及中醫藥材辭庫之研製;Design and Implementation of Medicinal Ontology and Thesaurus for Traditional Chinese Medicine
No full text[[abstract]]中醫藥從古至今已有二千多年歷史，由於經歷不同朝代的不同書寫及記錄方式，使得中醫藥之敘述豐富多樣。為了促進中醫藥的資訊化，本研究團隊首先進行中醫藥材之標準化，並建構一個中醫藥材知識本體。為了方便持續進行中醫藥材之標準化，本研究團隊也建構一個中醫藥材辭庫。中醫藥材知識本體及中醫藥材辭庫的開發是中醫藥資訊化的基礎建設。 本論文以本研究團隊先前建構的包含221個藥材功效、155味藥材與112帖方劑的中醫藥材辭庫為基底，新增了7個藥材功效、77味藥材與94帖方劑，主要是中醫實證相關的藥材功效、藥材與方劑。根據新增的藥材功效及藥材，使用本研究團隊開發的藥材配伍系統，新增的94帖方劑中，93帖方劑的所有藥材都能以首選藥材配伍出來。剩下1帖方劑為養生方劑，強調食物性藥材，而非藥材功效的效力，因此有一個藥材沒能以首選藥材配伍出來。另外，本論文也加強了本研究團隊先前建構的中醫藥材辭庫的查詢及維護功能。
Traditional Chinese Medicine (TCM) has developed more than 2000 years. The terms for description of medicinal effects exist high diversity. To promote applying information technology to TCM, our research team has initiated the standardization of TCM herbs and medicinal effects of TCM herbs. Our research team has also constructed a medicinal ontology and a medicinal thesaurus for TCM. The medicinal ontology and medicinal thesaurus for TCM are the infrastructure of applying information technology to TCM. This thesis extends our previous work, which includes 221 medicinal effects, 155 herbs, and 112 formulas, to 228 medicinal effects, 232 herbs, and 206 formulas. The increased 6 medicinal effects, 77 herbs, and 94 formulas are associated with excess syndromes in TCM. Using the medicinal combination system developed by our research team to combine these 94 new formulas, except for one herb in one formula, all herbs in 94 formulas are correctly recommended as the herb with the best effects. The failed formula is mainly for health care. Hence, food herbs with less effects rather than therapeutic herbs with better effects are adopted in that formula. In addition, this thesis also extends the query and maintenance functionalities of the medicinal ontology and medicinal thesaurus
基於Clang之程式碼檢查器設計與實作;A Security Checker Based on Clang
No full text[[abstract]]由於C語言在執行上的高效率、功能豐富、擴展性高、可移植性等優點，從被開發出來直至今2018年，其熱門程度一直與JAVA並駕齊驅，在軟體評價公司Tiobe程式語言排行榜中維持前兩名[1]。但是C語言本身的設計缺陷，卻也可能使得電腦當機、軟體崩潰，甚至是輕易的被駭客惡意入侵。這些漏洞包含了緩衝區溢位、格式化字串等等。 為了減少C語言漏洞對應用程式和作業系統的傷害，我們利用靜態分析工具來協助工程師去偵測原始碼中的漏洞。當被檢查的程式碼中使用了我們認為可能會造成系統危害的函式或變數時，編譯器會顯示相關的警告訊息，以便工程師能依據該訊息的提示，來得知原始碼中是否有漏洞的危險。 本篇論文開發出基於Clang之靜態分析工具來協助工程師檢查原始碼，利用檢查後輸出的警告提示，來讓程式執行的過程中擁有更高的安全性。
Due to the high efficiency, rich functions, high scalability, and portability of the C language, it has been developed since its inception in 2018, and its popularity has been keeping pace with JAVA, in the software evaluation company Tiobe programming language list. Maintain the top two [1]. However, the design flaws of the C language itself may cause the computer to crash, the software to crash, or even be easily maliciously invaded by hackers. These vulnerabilities include buffer overflows, formatted strings, and more.To reduce the harm of C language vulnerabilities to applications and operating systems, we use static analysis tools to assist engineers in detecting vulnerabilities in source code. When the function or variable that we think may cause system harm is used in the code being checked, the compiler will display a warning message so that the engineer can know whether there is a loophole in the source code according to the prompt of the message. Danger.This paper develops a static analysis tool based on Clang to assist engineers in checking the source code and using the warning prompts output after the check to make the program more secure during execution
基於編譯器偵測格式化字串攻擊;Compiler Support of Format String
No full text[[abstract]]由於C語言在執行上的高效率、功能豐富、可移植......等優點，從被開發出來一直到現在仍然維持在軟體評價網站中排行榜前兩名。並且C語言使很多程式撰寫者的程式啟蒙語言，因此相當多人在使用，但是由於C語言本身既有的漏洞，可能會導致這些軟替崩潰、當機、甚至會被攻擊者入侵。本篇論文主要希望可以解決C語言的程式漏洞:格式化字串攻擊，根據CVE的資料可以清楚地顯示出這一個至今仍然還是會發生，為了解決這一個問題本篇論文針對兩個議題提出辦法，第一:針對使用者所輸入的值去進行判斷，第二:去判斷字符數目跟參數數目是否一致，提出這兩個方法希望可以有效降低格式化字符攻擊的發生。此方法是透過共享函式庫的方法來達成我們希望達成的目的，讓程式可以不用重新進行編譯，並且可以在執行期進行管控，希望可以透過此方法使程式在執行的過程中有更高的安全性。
Due to the high efficiency, rich functionality, portability of the C language, it has been maintained in the top two rankings in the software evaluation website since its development. And the C language makes the program of many programmers enlighten the language, so a lot of people use it, but due to the existing vulnerabilities in the C language, it may cause these soft replacements to crashand even be attacked by attackers.This paper mainly hopes to solve the vulnerability of C language: format string attack, according to the CVE data, it can clearly show that this still happen. In order to solve this problem, this paper proposes two solutions. First, it is judged according to the input by the user. Secondly, it is determined whether the number of characters is consistent with the number of parameters, and the two methods are proposed to effectively reduce the occurrence of format string attacks.This method is to achieve the purpose by shared library, so that the program can be compiled without re-execution, and can be controlled during the execution period. We hope that the program can be made higher safety during the execution process
以空間、時間、轉換、及時空特徵作無參考視訊品質計算;No-reference Video Quality Metric Computation Using Spatial, Temporal, Transform, and Spatiotemporal Features
No full text[[abstract]]現今網路蓬勃發展，視訊提供者與使用者對於視訊品質的感知愈來愈重視，但受限於網路傳輸的頻寬，全參考、部分參考和無參考視訊品質計算中以無參考視訊品質評估為最佳的計算方法，其是最具彈性且廣泛使用的方法。本研究是基於無參考並提取空間、時間、轉換和時空特徵作為預測品質分數的依據。首先提取邊緣檢測和塊狀資訊做為空間特徵，再來提取亮度和運動狀態變化作為時間特徵。以提取離散餘弦轉換和小波轉換做中心點像素與周圍鄰居像素點的加強，並視為轉換特徵。考慮到可以同時提取空間和時間訊息，擷取軌跡的統計特性和三維離散餘弦轉換作為時空特徵。最後採用支持向量回歸預測最終品質分數。本實驗採用LIVE video quality assessment database 進行實作且實驗結果顯示該結果與其他現有方法相比具有較好的結果。
Nowadays, Internet is booming and the perception of video quality by video providers and users is becoming more important, but limit by the bandwidth of network transmission. No reference video quality computation is the best and well-known in three types of video quality assessment metrics. In this study, the proposed video quality computation metric is based on no reference and extracted spatial, temporal, transform, and spatiotemporal features as the basis for predicting quality scores. First, edge detection and blockiness are extracted as the spatial features and difference of luminance and motion are extracted as temporal features. The pairwise products of discrete cosine transform and wavelet transform are extracted to enhance the center point pixel and surrounding neighbor pixels, and are regarded as transform features. Considering that spatial and temporal information can extracted simultaneously, the statistical properties of trajectory and three-dimensional discrete cosine transform are taken as spatiotemporal features. Finally, support vector regression is utilized to predict the final quality score. This experiment using LIVE video quality assessment database and experimental results show that the results have better results than other existing metrics
英文版中醫症狀知識本體之研製;Design and Implementation of an English Version Symptom Ontology for Traditional Chinese Medicine
No full text[[abstract]]知識本體是某一領域相關概念的知識庫，包含這些概念具共識的術語及其定義，也包含概念與概念之間的關聯。一個領域知識本體的建構是該領域資訊化的基石。該領域的資訊系統可以透過知識本體自動分析及分享該領域具共識的知識。本研究團隊先前已進行中醫症狀的標準化，並研製一個中醫症狀知識本體。該症狀知識本體用四個性質來定義一個症狀：症狀物件、症狀屬性、症狀程度、症狀期間，這種方式可以定義大部分的中醫症狀。 本論文在中文版中醫症狀知識本體的基礎上，研製一個英文版中醫症狀知識本體。本論文在中醫症狀知識本體中，為每一個症狀、症狀物件、症狀屬性、症狀程度、及症狀期間，增加英文名稱及英文定義這兩個資料性質。運用中文名稱及英文名稱的對應，英文版中醫症狀知識本體的搜尋可以透過中文版中醫症狀知識本體的搜尋來完成。關鍵字: 中醫症狀標準化；中醫症狀知識本體；英文版中醫症狀知識本體。
An ontology is a knowledge base of common concepts in a domain, including terms and definitions of common concepts, as well as the relationships between concepts. The construction of an ontology for a domain is the cornerstone of applying information technology in that domain. Information systems for that domain can automatically infer and share knowledge about common concepts in that domain through the ontology. Our research team had conducted a standardization of symptoms in traditional Chinese medicine (TCM) and had developed an ontology for TCM symptoms. Symptoms in the ontology are defined using four attributes: symptom object, symptom property, symptom degree, symptom duration. This scheme can define most symptoms in TCM.This thesis develops an English version of the TCM symptom ontology based on the Chinese version of the TCM symptom ontology. This thesis extends two data properties: English name and English definition for each of the symptom object, symptom attribute, symptom degree, and symptom duration individuals. Making use of the mapping between the English name and the corresponding Chinese name, the queries in the English version of the TCM symptom ontology can be performed efficiently through the queries in the Chinese version of the TCM symptoms ontology.Keywords: standardization of TCM symptoms, TCM symptom ontology, English version of the TCM symptom ontolog
於低能見度之動態估算車輛安全距離;Dynamic Estimation of Vehicle Safety Distance under Low Visibility
No full text[[abstract]]追尾碰撞總會造成一些嚴重的傷害和財產損失，因此保持安全距離對於避免追尾碰撞是非常重要的。雷達碰撞警示系統常用於在即將發生追尾碰撞時去警告駕駛。然而，在惡劣天氣條件下，這種系統的精確度受到很大的限制。近年來，專用短距離通訊 (DSRC) 技術被提出來解決該問題。但是，仍然存在改進的空間，例如惡劣的天氣條件總是影響人的反應時間，並且如果這個警示系統沒有考慮到這一點，則它將無法去警告駕駛可能發生追尾碰撞。因此會影響追尾碰撞警示系統的準確性。為了增加追尾碰撞警示系統的精確度，我們提出了在低能見度動態估算車輛安全距離的系統。這個系統包含了感知反應時間模型，故障補償模型和在低能見安全距離模型。這個感知反應時間模型被用來估算在低能見度的反應時間，而這個故障補償模型被使用去補償DSRC傳輸延遲或故障時所遺失的資料，最後在低能見度安全距離模型被使用去估算一個最適合的警告距離。實驗結果表明，當存在DSRC故障或傳輸延遲時，我們可以估計可能的速度和車間距離。接下來，我們將所提出的方法與前車停止模型進行比較，其中平均警告距離比停止距離演算法多了5公尺。而在前車減速模型相比，我們提出的方法給出的平均警告距離比停止距離演算法多了1.8公尺。在能見度低的例子中，當能見度為50公尺時，平均警告距離比能見度為80公尺時的平均警告距離多了21.3公尺，這是因為能見度低造成我們的反應時間變長。而這些增加的距離讓系統能夠在車間距離位於警告距離和安全距離之間去警告駕駛，並且讓駕駛有足夠的距離去反應。
Rear-end collisions will always result in some serious injury and property damage, thus maintaining a safety distance is very important to avoid rear-end collisions. Radar collision warning systems are often used to alert drivers when a rear-end collision is about to occur. However, the accuracy of such systems is significantly limited in poor weather conditions. In recent years, Dedicated Short Range Communication (DSRC) technology was proposed to solve the problem. However, there is still room for improvement such as poor weather conditions always impact on the human reaction time, and if the warning systems do not take this into account, it will not be able to warn about a rear-end collision. The accuracy of the rear-end collision warning systems will thus be affected.For increasing the accuracy of the rear-end collision warning system, we propose a system for Dynamic Estimation of Vehicle Safety Distance under Low Visibility. This system includes aperception reaction time model, failure compensation model, and safety distance in low visibility model. The perception reaction time model is used to estimate reaction time in low visibility, the failure compensation model is used to compensate for DSRC transmission delays or failures, and the safety distance in low visibility model is used to estimate a most appropriate warning distance.Experiment results show that we can estimate the possible velocity and inter-vehicle distance when there are DSRC failures or transmission delay. Next we compare the proposed method with the Lead Vehicle Stopped (LVS) model, where the average warning distance is 5 meters more than that in the stopping distance algorithm. In comparison with the lead vehicle decelerating (LVD) model, the proposed method gives an average warning distance of 1.8 meters more than that in the stopping distance algorithm. In the case of low visibility, when the visibility is 50 m, the average warning distance is 21.3 m more than that of the case when the visibility is 80 m in both LVD and LVS, because low visibility leads to longer perception reaction times. These increased distances allow the system to warn the driver when the inter-vehicle distance is between the warning distance and the safety distance, and also allow the driver to have enough distance to react