京都大学 / 0048 / 新制・課程博士 / 博士(理学) / 甲第19885号 / 理博第4212号 / 新制||理||1605(附属図書館) / 32962 / 京都大学大学院理学研究科生物科学専攻 / (主査)教授 森 和俊, 教授 大野 睦人, 教授 杤尾 豪人 / 学位規則第4条第1項該当 / Doctor of Science / Kyoto University / DFAM

No description available.

在電腦視覺的研究之中，有關物件的偵測與追蹤應用在速度及可靠性上的追求一直是相當具有挑戰性的問題，而現階段發展以視覺為基礎互動式的應用，所使用到技術諸如:類神經網路、SVM及貝氏網路等。 本論文中我們持續深入此領域，並提出及發展一個方向性邊緣特徵集(DEM)與修正後的AdaBoost訓練演算法相互結合，期能有效提高物件偵測與識別的速度及準確性，在實際驗證中，我們將之應用於多種角度之人臉偵測，以及臉部表情識別等兩個主要問題之上；在人臉偵測的應用中，我們使用CMU的臉部資料庫並與Viola & Jones方法進行分析比較，在準確率上，我們的方法擁有79% 的recall及90% 的precision，而Viola & ones的方法則分別為81%及77%；在運算速度上，同樣處理512x384的影像，相較於Viola & Jones需時132ms，我們提出的方法則有較佳的82ms。 此外，於表情識別的應用中，我們結合運用Component-based及Action-unit model 兩種方法。前者的優勢在於提供臉部細節特徵的定位及追蹤變化，後者主要功用則為進行情緒表情的分類。我們對於四種不同情緒表情的辨識準確度如下：高興(83.6%)、傷心(72.7%)、驚訝(80%) 、生氣(78.1%)。在實驗中，可以發現生氣及傷心兩種情緒較難區分，而高興與驚訝則較易識別。 / Rapid and robust detection and tracking of objects is a challenging problem in computer vision research. Techniques such as artificial neural networks, support vector machine and Bayesian networks have been developed to enable interactive vision-based applications. In this thesis, we tackle this issue by devising a novel feature descriptor named directional edge maps (DEM). When combined with a modified AdaBoost training algorithm, the proposed descriptor can produce effective results in many object detection and recognition tasks. We have applied the newly developed method to two important object recognition problems, namely, face detection and facial expression recognition. The DEM-based methodology conceived in this thesis is capable of detecting faces of multiple views. To test the efficacy of our face detection mechanism, we have performed a comparative analysis with the Viola and Jones algorithm using Carnegie Mellon University face database. The recall and precision using our approach is 79% and 90%, respectively, compared to 81% and 77% using Viola and Jones algorithm. Our algorithm is also more efficient, requiring only 82 ms (compared to 132 ms by Viola and Jones) for processing a 512x384 image. To achieve robust facial expression recognition, we have combined component-based methods and action-unit model-based approaches. The component-based method is mainly utilized to locate important facial features and track their deformations. Action-unit model-based approach is then employed to carry out expression recognition. The accuracy of classifying different emotion type is as follows: happiness 83.6%, sadness 72.7%, surprise 80%, and anger 78.1%. It turns out that anger and sadness are more difficult to distinguish, whereas happiness and surprise expression have higher recognition rates.

區間型資料一般定義為由一個連續型變數的上限及下限所構成，本文中，我們特別引進了一種與時間相關的區間型資料，在Irpino (2006, Pattern Recognition Letters, 27, 504-513)，他提出每個觀測值皆是由某個時段的起始值及終點值之有方向性的區間所組成，譬如某一支股票在某一周的開盤價和收盤價。過去已經有許多方法運用在區間型資料，但尚未有方法來處理有方向性的區間型資料，然而Irpino 延伸主成分方法來處理有方向性的區間資料。Irpino提出的方法以幾何學的觀點來說，可視為定義在多維度空間上對有方向性線段(一般都稱作“spaghetti”)的分析，在本文中我們有更作進一步的延伸，不僅引入股票的開盤價及收盤價，且引入當周的最高價及最低價來探索Irpino所遺漏的資訊。此外，我們也嘗試用貝他分配來取代Irpino所使用的均勻分配來檢測是否有明顯的改善。延伸的方法需要計算大量複雜的式子，包含了平均數，變異數，共變異數等，最後利用相關係數矩陣進行主成分分析。然而最後的結論為若考慮資訊的價值，以加入最高值和最小值的延伸方法是較好的選擇。 / Interval data are generally defined by the upper and the lower value assumed by a unit for a continuous variable. In this study, we introduce a special type of interval description depending on time. The original idea (Irpino, 2006, Pattern Recognition Letters, 27, 504-513) is that each observation is characterized by an oriented interval of values with a starting and a closing value for each period of observation: for example, the beginning and the closing price of a stock in a week. Several factorial methods have been discovered in order to treat interval data, but not yet for oriented intervals. Irpino presented an extension of principle component analysis to time dependent interval data, or, in general, to oriented intervals. From a geometrical point of view, the proposed approach can be considered as an analysis of oriented segments (nicely called “spaghetti”) defined in a multidimensional space identified by periods. In this paper, we make further extension not only provide the opening and the closing value but also the highest and the lowest value in a week to find out more information that cannot simply obtained from the original idea. Besides, we also use beta distribution to see if there is any improvement corresponding to the original ones. After we make these extensions, many complicated computations should be calculated such as the mean, variance, covariance in order to obtain correlation matrix for PCA. With regard to the value of information, the extended idea with the highest prices and the lowest price is the best choice.

送氣與非送氣的對比是台灣華語塞音及塞擦音之特性。為討論送氣的議題，本篇論文透過語音實驗，從兩方面來探討：其一為嗓音起始時間(voice onset time)，另一則為語誤的方向性(directionality)及標記 (markedness)。 本研究有兩個實驗，皆以四字非詞為實驗材料。實驗一為語誤實驗，此實驗結合繞口令以及立即回想的方法來誘發語誤；實驗二則為嗓音起始時間實驗，只讓受試者回想，並未結合繞口令。這兩個實驗的設計由於受到中文本身音韻限制及詞彙空缺的限制，韻母使用了單母音(V)及韻母(VG或GV)兩種形式，因此實驗二亦將母音對子音數值的影響之比較結果納入討論。 嗓音起始時間實驗的結果顯示，使用四字非詞並結合回想作為實驗內容所測量出的數值與前人使用實詞所測量出的數值並無太大不同，並支持送氣子音的數值較非送氣子音的大。此外，塞音的研究結果支持Cho & Ladefoged (1999)所提出的原則，也就是塞音除阻的位置若越後面，則其數值會越大；若將塞音的數值套用Cho & Ladefoged (1999)的分類，則中文的非送氣塞音確為非送氣子音，而送氣塞音的數值則介於稍微送氣及送氣子音之間。塞擦音的研究結果若如Lai (2013)般套用Cho & Ladefoged (1999)之原則，則會發現送氣塞擦音符合此原則，但非送氣塞擦音卻非如此。另外，子音後接的母音或韻母亦會影響子音之嗓音起始時間。研究結果顯示當大部份子音後接單母音時，子音之嗓音起始時間較長；研究結果亦指出當子音後接的為單母音[a]或以[a]開頭的韻母時，子音的嗓音起始時間會較接其他母音如[u], [i], [ow]時短。 語誤實驗的結果顯示，除了[k], [kh]有以有標送氣塞音取代無標非送氣塞音的情形外，其他組塞音及所有塞擦音皆無以有標送氣子音取代無標非送氣子音(反之亦然)的情形。就方向性而言，研究結果顯示語誤來源出現在語誤之前(perseveration)的方向性顯著最多。 / The contrast of aspiration is one of the characteristics of stops and affricates in Taiwan Mandarin. Through the conduction of two phonological experiments, we discuss the issue of aspiration from two aspects: one is from voice onset time (VOT), and the other one from directionality and markedness of speech errors. Non-word quadruple sets were used in both two experiments. The first experiment was speech errors, which combined tongue twister and immediate recall to induce speech errors; the second experiment was voice onset time, which only had the subjects to recall. Due to the limitation of phonological ill-form and lexical gaps, the rhymes of the two experiments had two forms, V and VG/GV; as a result, how vowels affected the VOT of stops and affricates were also discussed. The findings of the second experiment (voice onset time) showed that the VOT measured by using non-word quadruple set was similar to the VOT measured by using real words as stimuli, and the findings also indicated that the VOT of the aspirated consonants was longer than that of the unaspirated ones. In terms of the results of the stops, they supported the principle proposed by Cho & Ladefoged (1999) that the duration of VOT has a further back relationship with the closure; furthermore, the categories proposed by Cho & Ladefoged (1999) were applied onto the results, it showed that Mandarin unaspirated stops were indeed unaspirated consonants, and the aspirated stops fell between slightly aspirated and aspirated consonants. The findings of the aspirated stops, if applied the principle proposed by Cho & Ladefoged (1999) like Lai (2013), revealed that aspirated affricates supported the principle, yet the unaspirated ones did not. Regarding the influence of vowels on the VOT, it was found that when the succeeding vowel was a single vowel [a] or rhymes beginning with [a], the VOT of the stops and the affricates would be shorter. The findings of the first experiment (speech errors) indicated that with regard to the stops, the aspirated velar stop tended to replace the unaspirated counterpart, while for the other stops, the probability for marked aspirated stops to replace unmarked unaspirated ones or vice versa was of the same. With regard to the directionality of speech errors, the findings showed that perseverations significantly outnumbered anticipations and exchange.

全球暖化已成為大多數國家近幾年所關心的議題。雖然臺灣政府於2010年5月所核定的「國家節能減碳總計畫」之中，儘管減量目標及原則業已確立，但未充分探討各別產業的面臨碳排放管制時所造成的影響，因此，臺灣各產業或次部門究竟應承擔多大的減量責任仍是混沌不明。 為了更清楚瞭解碳排放管制對於臺灣工業部門之各產業的影響，本文以方向性距離函數，估算工業部門中14個產業367家上市櫃、興櫃和公開發行公司於2005年至2010年，在不同電力消費所產生的CO2之責任歸屬情況下之生產效率及汙染減量成本，以反映各產業於管制下的機會成本，便於鎖定某些產業或公司，來賦予減量責任。結果發現，各產業在實施碳排放管制後的效率水準，會高於未實施碳排放管制時的效率水準，且各產業的平均總汙染減量成本與每噸二氧化碳減量成本相差甚大。而臺灣尚未通過相關法規以規範各產業二氧化碳排放水準，若各產業節能技術或政府相關配套政策未改善，倉促實施碳排放管制，除了對管制對象的產生影響外，也可能間接衝擊未管制對象。政府除了加強輔導各種產業從事節能技術外，也需要適當的公布各產業各公司的各種汙染排放量資訊，以供各界研究碳排放管制或其他汙染排放管制對社會的影響。 / Global warming has become the topic of most countries which concerns things in recent years. Government sets up CO2 reduction objectives and principle in “General National Plan for Energy Saving and Carbon Reduction” in 2010, but the plan doesn’t probe effect of CO2 regulation which different industries. Therefore, it is unknown that different industries should be responsible for the abatement of CO2. In order to clearly understand effect of CO2 regulation for Taiwan industrial sector, this paper use directional distance function to estimate production efficiency and pollution abatement cost of 367 public companies in 14 industries in industrial sector from 2005 to 2010, and to reflect different industries’ opportunity cost under CO2 regulation. Production efficiency of different industries after the implementation of CO2 regulation will be higher than before the implementation of CO2 regulation. On the other hand, different industries have a variety of average pollution abatement cost and pollution abatement cost of CO2 per ton. However, Taiwan has not yet adopted the environment laws to set up industrial CO2 emission level, if the government hurriedly implemented CO2 regulation for Taiwan industrial sector, may be indirectly influence other sectors. Hence, the government should not only urge that industries must be engaged in energy-saving technologies, but also announce companies’ various pollution emission information which provides research institutes to analyze effect of social welfare under CO2 regulation.