本論文應用攝影構圖法則，以模糊邏輯理論為基礎，判別影像的攝影構圖類型。構圖(Composition)乃是攝影這項平面藝術創作最重要的美學元素之一，其目的是利用空間中的物體配置，經由透視投影後，讓畫面的整體呈現平衡感；專業的、優秀的攝影作品，皆會符合攝影的基本構圖原理。因此許多的影像增強、影像合成的應用中，也應該配合相片原本的構圖設計，針對所欲表達的重點予與適當地調整，而非「盲目的」以一體適用的法則去處理每一張照片。 論文中，我們針對影像所欲表達的重點區域，分析其結構特性，設計不同的特徵，並以模糊邏輯理論為基礎，應用Mamdani系統，結合隸屬函數與攝影構圖判別法則的交互作用，用以辨認所欲處理相片的構圖類別。依據辨認後的構圖類別，即可對該影像做適當地分割及調整，以使相片能有最佳的影像增強處理。 實驗證明，本文所提出的方法能有效地辨認攝影構圖類別，針對不同攝影構圖所作的影像修正，才能更符合人眼的視覺喜好。 / This thesis addresses the problem of how to recognize the photographic composition from a given photo based on the theory of fuzzy logic. Composition is one of the important aesthetics for the plane figure photo art. To present the balance of its holistic picture, it takes the advantage of special object arrangement after acting perspective projection. A piece of professional and qualified photo work will realize these basic photo composition methods. For many applications about the digital photo, the operations, i.e., photo enhancement, segmentation, output, and synthesis, all need to match up the photographic composition to do accurate processing rather than “blind” processing that assumes each photo with the same “composition.” An automatic recognition method using image features from some specific regions is described. The method is employed in a Mamdani model and combines outputs of multiple fuzzy logic rules and feature extraction algorithms to obtain confidences that can identify the correct photographic composition. Experimental results show that the proposed method is robust and effective for photographic composition recognition. The feature with adjusting in different photo composing will be able to comfort our human sight.

本研究探討的個案公司為一間自動辨識類產品供應商，總部位於美國，在台設有子公司從事製造與亞太區業務開發，其供應之零件在POS系統產業供應鏈中具有相當的知名度。本研究採取個案分析法，探討個案公司在海外市場進入策略中，如何選定日本為進入市場、並透過何種方式進入的策略。同時，利用波特的五力分析原型探討個案公司在日本市場所呈現的企業競爭力。此外，採用邱志聖(2010)策略行銷分析架構來分析進入日本市場後，透過在地代理商作為居中的角色，如何在個案公司與日本終端客戶之間互動，來降低個案公司與終端客戶之間交易行為產生的四種成本，其四種成本分別為：外顯單位效益成本、資訊搜尋成本、道德危機成本、專屬陷入成本。最後，再透過個案的觀察與分析的結果，提出對個案公司的未來經營日本市場以及與代理商合作關係的建議。 本研究主要發現為，個案公司為了使企業內部資源運用最大化，並向外發展利基市場，因此選擇進入日本市場。基於風險評估及交易成本的考量，個案公司發展出當地代理制度與日本當地具產業經驗者合作來推行個案公司專案的服務。 透過代理制度引進後，終端客戶與個案公司之間的交易成本雖然在外顯單位效益成本上的變化無法做明確預估之外，在資訊搜尋、道德風險、及專屬陷入成本上，都有顯著下降，且個案公司的日本市場的整體營收有明顯成長。 本研究對個案公司提出的建議有三點，分別為：一、持續適度地參與代理商與終端客戶的接觸，協助降低終端客戶的交易成本並提升合作信心。二、增加代理商的專屬陷入成本，降低其外顯效益成本，以保持代理合作的意願及穩定性。三、若日本代理商因各種可能原因無意願繼續代理時，對日本市場需有配套措施。 / This study aims to explore the case story of a U.S. based company in automatic identification industry, which has a manufacture plant and sales team in Taiwan. The study is done by a case study method, to analyze the case company’s foreign market entry strategy, and how the decision is made to choose Japan as the target market as well as which entry strategy is performed. Porter’s Five Force Model is used in the analysis to determine the case company’s competitive power in Japan market. In addition, by setting up distribution system, the costs including cost of utility, information research, moral hazard, and asset specify to the end customer in dealing with case company has been influenced dynamically. And based on the result and observation of the study, then suggestions will be provided to the case company as conclusion. Findings of the study are that the decision on how to enter Japan market is made by considering the overall all risks and transaction costs among several entry modes. Then the distribution system is implemented, and a good market position along with the market trend helped the case company find customers with right scale in niche market, and it is evident that the distribution system did lower down the 4 kind of costs to the end customer in general, and the overall market revenue increased in Japan as return. In conclusion, three suggestions are made to the case company: 1. Involve with distributor and increase the interfaces with the end customers to strengthen the case company and distributor’s credibility and also the confidence of end customer. 2. Tighten the cooperation with distributor by increasing distributor’s cost of specify to ensure the continuous partnership. 3. To keep the option of direct managing Japanese account in the near future in case of the distribution partnership cannot be continued.

特徵描述子為電腦視覺中相當重要的一部分，本論文基於知名的特徵描述子：區域二元化圖型的架構上，提出了新的特徵描述子，並將其命名為延展式區域三元化圖型。我們所提出的特徵描述子與區域二元化圖型相比，有著較強的抗噪能力而且保留了區域二元化圖型簡單的計算複雜度。本論文也探討了區域三元化圖型中是否存在著uniform pattern，基於區域二元化圖型中uniform pattern的定義，我們提出了屬於區域三元化圖型的uniform pattern，並在圖像實驗中驗證了其大量存在性。我們將區域三元化圖型應用於材質分析與人臉辨識中，實驗結果驗證了本論文所提出的特徵描述法在這些應用的優越性。 / Robust feature descriptor is essential in developing effective computer vision applications. In this thesis, we present an extension to the well-known local binary pattern (LBP) feature descriptor. The newly defined descriptor known as extended local ternary pattern (ELTP) exhibits better noise resistivity than the original LBP, while maintaining computational simplicity. We further investigate the presence of uniform patterns in ELTP. With a slight modification of the definition of uniformity, it is found experimentally that uniform ELTPs account for 80% of all patterns in texture images. The proposed ELTP and uniform ELTP are applied to object classification tasks, including texture analysis and face recognition. Experimental results validate the superiority of ELTP over conventional LBP approaches.

隨著科技的進步，網際網路已充斥在我們的生活之中。音樂也不再以硬體儲存的方式流傳（例如ＣＤ、黑膠唱片），而是轉變為數位音樂的方式，透由網路平台散播。許多數位音樂串流服務平台網站也如雨後春筍般誕生，例如iTunes、Spotify、Musicovery。加上文化水平的提升，音樂已是現代人生活之中，不可或缺的一部分。世界上的音樂難以計數，如何將音樂分門別類做好管理乃為現代商業應用的一個重要課題。因此，音樂曲風自動化辨識的技術確實為一個實用且難以迴避的課題。 過去在曲風自動化辨識已有許多研究，但內容不外乎音訊處理、頻譜轉換、特徵擷取、特徵降維、監督式學習機。在相同的模式下提出各種改良，或是全新的特徵擷取…諸如此類，而辨識率也達到了七成以上。本篇論文採用不同於以往的做法，將訊號進行頻譜轉換後層層降維，所得之訊號搭配LDA與決策樹進行辨識，最後去比較與分析離散餘弦轉換與小波轉換在辨識率上的優劣。我們發現搭配小波轉換與混合LDA及決策樹的方法，可以將音樂曲風之分辨率達到八成五以上。

中國有句諺語:「民以食為天」，食物乃人類生活所不可缺的要素之一，而人們對於食物則有各自的偏好，而要從琳瑯滿目的食物中依照個人喜好來推薦則成為一門重要的課題。 隨著科技的進步，智慧型手機的出現為人類帶來了許多便利，也逐漸改變了人們的生活方式，群眾可以透過智慧型手機來記錄生活的點滴，記錄的方式正走向數位化，而如何利用這些累積下來的數位資料來做分析與推薦也成為熱門的研究目標。 本論文從味覺方面著手，將LifeLog的飲食記錄與味覺做結合，並透過大眾分類與群眾外包的方式，將味覺資料由智慧型手機使用者處獲得，並建構成包含餐廳、餐點名稱以及其對應味覺之資料庫。 本論文實作了一個程式Foodtaste，包含了記錄餐點味覺資料，查詢個人記錄，以及實作數種推薦的功能。本論文並提出了數個計算方法，透過LifeLog累積下來的味覺資料進行計算，來獲得每位使用者的個人口味偏好和味覺比例，並將這些資料與餐點的味覺比例計算來對餐廳進行個人化的餐點推薦。 / Foods and eating are the basic element of human's life, and people have their own favorite in choosing foods. Thus it is an important issue to make some recommendation for people in front of a dazzling array of foods. With the advances in technology, smartphones bring convenience to people and change their life style. One can use smartphones to record various things in his life. The ways of memories become digitalized, and how to use these digital data to analyze and give opinions becomes popular. Base on one’s taste, present study combined dietary records and food taste in Lifelog, using Folksonomy and Crowdsourcing to acquire data of specific food taste from smartphone users, and linked these data to restaurant’s name and the name of the meal in our database. We designed a smartphone application which called "Foodtaste". It provided users to record what they ate and how did it taste, looking up personal records, and several recommending methods. Our study also provides several methods in calculating cumulative data in Lifelog and acquiring the preference of one’s taste and ratio in variable foods from every user. Then we calculated these data to carry out personalized food recommendation.

本研究以事件相關電位方法,探討中文字在進行聽覺詞彙辨識的作業中,分別受字形表徵的同音字密度多寡、以及字音-字形對應一致性的影響。首先 進行中文字的字音-字形一致性語料庫的建立,量化中文字的音形對應一致性程度。透過明確的定義並操弄中文字的音形對應一致性,以及中文字同音 字密度的特性,探討在進行中文字聽覺詞彙辨識歷程中,字形表徵屬性如何對語音辨識作業產生影響的認知歷程。實驗分為三種情境(1)同音密度低 (low HD)、(2)同音密度高/音形對應一致性高(high HD/high P-O)、 (3)同音密度高/音形對應一致性低(high HD/low P-O)。前兩項的比較為高同音字密度下的音形對應一致性效果,而後兩項的比較則為在高音形對應一致性下的同音字密度的效果。實驗一採語意判斷作業,研究結果顯示,在高同音字密度時,高音形對應一致性的字引發較大的 N400;而在高音形對應一致性時,同音字密度效果在 LPC 得到顯著的差異。反應中文字在聽 覺詞彙判斷作業上,字形可自動被激活,進而影響語音的辨識。實驗二採押 韻判斷作業,研究結果同樣發現高音形對應一致字引發較大的 N400,實但同音字密度效果性效果在判斷押韻作業上並未達顯著效果。另外,在高音形對應一致性情形下所得到的押韻效果最大,尤以同音字密度高且音形對應一致性高的情況下,押韻效果出現的時間較早。本研究結果支持雙向交互激發模型(BIAM)的假設:中文口語詞彙辨識的歷程中,也會自動激發字形訊息,而語音、字形、語意之間的對應一致性越好,口語詞彙辨識及其整體處理的效能也越高。

創業之目的是突破現狀，追求創新，創業家必須擁有資源才能創新產品與服務，但他們卻因資源受限而有志難伸。新創企業如何以拼湊有限資源來翻轉劣勢，是隨創（bricolage）所提倡的主導做法。可是，拼湊資源是透過怎樣的時機，當今文獻卻討論有限，因此，時機辨識成為是當前亟需探索之重點。本研究選取三個原住民族新創企業個案，分析他們的經營歷程、挑戰與如何透過辨識時機進行資源拼湊。本研究發現，原住民族文化是重要的資源，這些在地資源成為識別機會、突破資源稀缺的服務創新關鍵。在學術意涵上，本研究提出三個議題。第一，辨識時機需要觀察資源轉換脈絡；第二，辨識時機需要設定資源使用情境；第三，辨識時機需找出資源調合方式。在實務啟示上，本研究提出辨識時機的各種可能，並歸納多樣的隨創原則。進行隨創時，若能掌握在地文化便可以找出適當時機，讓有限的資源拼湊出無限的創新。 / The objective of entrepreneurship is to seek breakthrough innovation within status quo. Entrepreneurs must obtain resources in order to innovate new products or services. But their aspirations often are frustrated due to resource constraints. The dominant approach to entrepreneurial bricolage is to advise new ventures to turnaround adversities by making-do with resources. But, when could entrepreneurs identify opportunity for tinkering limited resources? This is an inquiry worth further investigation. This research selects three case studies of indigenous start-ups while analyzes their entrepreneurial process and challenges within; and examines how they recognize opportunities in creative resource tinkering. This study found that indigenous culture is a kind of critical local resource, which may trigger opportunity recognition in order to guide resource making-do and achieve service innovation. Theoretically, this study suggests that (1) opportunity recognition must understand the context that is suitable for the transformation of resource quality; (2) opportunity recognition must pay attention to resource deployment in situ; (3) opportunity recognition should identify methods of resource blending. Practically, this study proposes several possibilities of opportunity recognition and summarizes a variety of bricolage principles. When engaging entrepreneurial bricolage, if we could get a grip with local culture, it increases our chance to identify the right moment and allow limited resources to enact unlimited innovation.

大陸及臺灣均規定，發明專利權人專有排除他人未經其同意而實施該發明之權。亦規定，物之發明之實施，指製造、為販賣之要約、販賣、使用或為上述目的而進口該物之行為。在專利權人或其專屬被授權人提起專利侵權實體訴訟前，專利權人或其專屬被授權人一般得請求海關就嫌疑貨物為暫不放行措施或扣押，以防止侵害專利權之貨物進入商業市場。一般而言，涉及商標權或著作權之貨物較容易透過貨物外觀以觀察是否侵權。然而，大部分附有專利的貨物均不易以貨物外觀以觀察是否侵權。特別是，舉例而言，一支智慧型手機上可能有上千的專利於其上，而且許多商品都有類似情形。所以，由第一線的海關人員去辨識有無專利侵權是困難的。 「與貿易有關的智慧財產權協定」第52條規定，任何權利人申請第51條之暫不放行措施，都應向主管機關提出足以推定在進口國法律之下有侵害權利持有人智慧財產權之表面證據，並就有關物品提供詳細說明，俾使海關易於辨認。如果權利持有人提供有侵害其智慧財產權之表面證據，仿冒品或是盜版品一般來說較易於由海關第一線人員透過貨物外觀加以辨識。另一方面，專利權人或其專屬被授權人提出侵害其專利權之表面證據是困難的。 除上述問題之外，尚有一更困難的問題有待專利權人或其專屬被授權人解決。那便是，如果專利權人或其專屬被授權人不指派間諜至競爭對手之中，並不知道涉及侵害專利權之貨物何時、何地將會進口。因此，本論文嘗試研究如何扣押侵害專利權貨物之議題。 本論文區分為六個部分。第一章是緒論，說明研究動機及目的，以及研究方法及範圍。第二章是說明及定義智慧財產權邊境執行之指導原則，以及第一線海關人員依據貨物外觀辨識原則查尋嫌疑貨物。第三章是討論智慧財產權邊境執行之三種方式。第四章是分析與比較兩岸專利權之執行。第五章是闡述司法暫時權利保護與專利權邊境執行之配合。第六章是結論。 / Both Mainland China and Taiwan enacted laws to protect patentees’ exclusive rights to the invention and to prevent the rights from being exploited, without the patentee’s consent, via making, offering for sale, selling, using or importing of the infringed goods. Prior to filing the infringement litigation, the patentee or the exclusive licensee may generally initiate to request its customs authorities to suspend the release of, or to detain, suspect goods that involve the infringement of a patent from entering into the channels of commerce. Generally speaking, goods that involve trademark or copyright can be more easily observed whether it is infringed via its apprearance. However, most of goods that involve patents can not be more easily observed whether it is infringed, and inter alia, for example, there are probably more than 1,000 patent rights in one smart phone, and many goods have the similar situations. So it is difficult to distinguish infringed or non-infringed goods by front line staffs of customs. Article 52 of Agreement on Trade-Related Aspects of Intellectual Property Rights defines suspension of release. It provides: “Any right holder initiating the procedures under Article 51 shall be required to provide adequate evidence to satisfy the competent authorities that, under the laws of the country of importation, there is prima facie an infringement of the right holder’s intellectual property right and to supply a sufficiently detailed description of the goods to make them readily recognizable by the customs authorities.” It’s understood that the counterfeit trademark or pirated copyright goods can be generally observed via its appearance by front line staffs of customs, if a right holder supply prima facie evidence of infringement of the right holder’s intellectual property right. On the other hand, it is very difficult for a patentee or a exclusive licensee to supply prima facie evidence of infringement of the patent. In addition to the above mentioned issues, there is a more severe problem that the patentee or the exclusive licensee needs to solve. It’s that the patentee or the exclusive licensee doesn’t know when and where the goods that involve the infringement of patents will import, if the patentee or the exclusive licensee seek for the goods that involve the infringement of patents without designating spies in rivals. Therefore, this thesis tries to research the issue with respect to how to detect and detain goods that infringe patent rights. This thesis is divided in six parts. Chapter 1 makes a description of the motive and the purpose of this article. It also includes the method and the range of this research. Chapter 2 explains and defines “The Guidelines of Intellectual Property Rights Border Enforcement” and the suspect of goods could be found by front line staffs of customs in compliance with “The Principle of Goods Appearance Identification”. Chapter 3 discusses three modes of intellectual property rights enforcement. Chapter 4 analyzes and compares Cross-Strait Customs in patent border enforcement. Chapter 5 describes cooperation of provisional measures and patent border enforcement. Finally, Chapter 6 is the conclusion.

貼心的智慧型生活環境，必須能在不同的情緒狀態提供適當服務，因此我們希望能開發出一個情緒辨識系統，透過對於形於外的生理感測資料的變化來觀察形於內的情緒狀態。 首先我們採用國際情緒圖庫系統(IAPS: International Affective Picture System) 及維度式分析方法，透過心理實驗的操弄，收集了20位的受測者生理數值與主觀評定情緒的強度與正負向。我們提出了一個情緒辨識學習演算法，經由交叉驗證訓練出每個情緒的特徵，並藉由即時測試資料來修正情緒特徵的個人化，經由學習趨勢的評估，準確率有明顯提升。其次，我們更進一步引用了維度式與類別式情緒的轉換概念來驗證受測者主觀評定的結果。相較於相關研究實驗結果，我們在維度式上的強度與正負向辨識率有較高的表現，在類別式上的驗證我們也達到明顯區分效果。 更重要的是，我們所實作出的系統，是搭載了無線生理感測器，使用時更具行動性，而且可即時反映情緒，提供線上智慧型服務。 / A living smart environment should be able to provide thoughtful services by considering different states of emotions. The goal of our research is to develop an emotion recognition system which can detect the internal emotion states from external varieties of physiological data. First we applied the dimensional analysis approach and adopted IAPS (International Affective Picture System) to manipulate psychological experiments. We collected physiological data and subjective ratings for arousal and valence from 20 subjects. We proposed an emotion recognition learning algorithm. It would extract each pattern of emotions from cross validation training and can further learn adaptively by feeding personalized testing data. We measured the learning trend of each subject. The recognition rate reveals incremental enhancement. Furthermore, we adopted a dimensional to discrete emotion transforming concept for validating the subjective rating. Compared to the experiment results of related works, our system outperforms both in dimensional and discrete analyses. Most importantly, the system is implemented based on wireless physiological sensors for mobile usage. This system can reflect the image of emotion states in order to provide on-line smart services.

本論文使用非監督式新細胞認知機（Unsupervised neocognitron）神經網路來便是印刷體中文字。 關於非監督式新細胞認知機，本論文提出兩項修改。第一項，Us1子層的結點不進行學習，而是直接套用人為方式所指定的12個區域特徵，而Us1之後的S子層仍然使用非監督式學習的方式決定其所要偵測的區域特徵。第二項修改則是，在學習過中設定一個上限值來限制代表節點（representative）產生的個數。如此設計的目的是為了避免模板（cell-planes）分配不均的問題。在本研究，採用這兩項修改的新細胞認知機稱為模式一，而使用第二項修改的新細胞認知機稱為模式二。 本論文裡的所有實驗分為兩部分。在第一部分有四個實驗，這些實驗都使用相同的訓練範例與測試範例。訓練範例有兩組，第一組包含“川”,“三”,“大”,“人”,“台”等五個中文字。而第二組包含“零”,“壹”,“貳”,“參”,“肆”等中文字。訓練範例都市採用細明體，而測試範例則是採用其他九種不同字體。第一個實驗的主要目的是測試模式一的績效。實驗結果顯示，模式一很容易學習成功而且辨識率可以接受。另外三個實驗的目的是想要了解某些參數值與系統績效的關係。這些參數包含S-欄的大小（the size of S-column），模板樹（the number of cell-planes），以及節點的接收場大小（the size of cells’ receptive field）。這三個實驗所使用的網路系統是模式一。 第二部分有二個實驗，主要的目的是比較模式一與模式二的系統績效。在第一個實驗，所使用的訓練範例與第一部分實驗相同。實驗結果顯示模式一比較容易成功地學習，而且系統有不錯的表現。第二個實驗，使用17個中文字做為訓練範例。這17個字包括“零”,“壹”,“貳”,“參”,“肆”,“伍”,“陸”,“柒”,“捌”,“玖”,“拾”,“佰”,“仟”,“萬”,“億”,“圓”,“角”。實驗結果顯示，模式一仍然是一個不錯的系統。 / In this study, we are investigating the feasibility of applying the unsupervised neocognitron to the recognition of printed Chinese characters. Two propositions for the unsupervised neocognitron are mentioned. The first on proposes that the input connections of the first layer are manually given, and all subsequent layers are trained unsupervised. The second one concerns the selection of representatives. During the process of learning, the number of cell-planes that send representatives for each training pattern has an upper bound. The unsupervised neocognitron with implementing these two propositions is named as Model 1, and the unsupervised neocognitron with implementing only the second proposition is named as Model 2. Experiment in this study are grouped into two parts, called Part I and Part II. In Part I, four experiments are conducted. For each experiment, two sets of training patterns will be conducted respectively. The first one, called the simple training set, consists of five printed Chinese characters“川”,“三”,“大”,“人”, and “台” with size of 25*25 in MingLight font. The second one, called the complex training set, contains another five printed Chinese characters“零”,“壹”,“貳”,“參”, and “肆” in the some font and size. After training, these characters of other nine different fonts are presented to test the generalization of the network. The objective of the first experiment of Part I is to investigate the performance of Model 1. Simulation results shot that Model 1 demonstrates a good ability to achieve a successful learning. In other three experiments, the effect of choosing different value for some parameters in investigated. The parameters include the size of S-column, the number of cell-planes, and the receptive field of cells. In Part II, a comparison of the performance of Model 1 and Model 2 is made. In the first experiment, Model 1 and Model 2 are trained to recognize the simple and complex training sets described above. Experimental results show that Model 1 shows higher ability to achieve a successful learning, and performance of Model 1 is acceptable. In the second experiment, 17 training patterns are presented during the learning process. These training patterns include “零”,“壹”,“貳”,“參”,“肆”,“伍”,“陸”,“柒”,“捌”,“玖”,“拾”,“佰”,“仟”,“萬”,“億”,“圓”,, and “角”. From the simulation results, Model 1 is a promising approach for the recognition of printed Chinese characters.