1 |
基於歌詞文本分析技術探討音樂情緒辨識之方法研究 / Exploring Music Emotion Recognition via Textual Analysis on Song Lyrics陳禔多 Unknown Date (has links)
音樂是一種情感豐富的媒體。即使跨越了數個世紀,人們還是會
對同一首歌曲的情緒表達有類似的理解。然而在現今的數位音樂資料
庫可以看出,我們是不可能憑著人力完成數量如此龐大的音樂情緒辨
識,也因此期待電腦可以協助完成如此繁重的工作。隨著機器學習的
發展,電腦逐漸可以透過統計模型與數學模型判斷與辨識一些並未事
先提供規則的資料,而無法言傳的音樂情緒也得以有機會交由電腦辨
識、分類。雖然目前有許多透過訊號處理技術進行的音樂辨識研究,
但是透過歌詞文本的辨識卻是相對少見,使用的特徵也多侷限於通用
的文字資訊。本研究以音訊特徵為基礎,從不同的歌詞文本資訊出
發,透過分析歌詞文本進行歌曲情緒辨識,提供更多優化的參考資
訊,藉以提升歌曲於交流、表達、推薦等互動的功能性與準確性。實
驗結果發現,歌詞文本資訊對於歌曲的正負面情緒辨識確實有相當好
的表現,而對於特定分類的限制則是值得更多透過不同自然語言處理
的方法強化的。
|
2 |
人機介面中的形狀辨識及其應用 / Shape recognition and its application in human-computer interaction鄭聖耀, Cheng, Sheng Yao Unknown Date (has links)
電腦硬體的發展日新月異,電腦在運算的能力有長足的提升,遠遠超過一般人腦的計算能力,隨著電腦的普及率大幅提高,電腦由以往為專業領域的工具轉變為家庭不可或缺的商品,一般大眾也成為電腦的使用者,與電腦溝通的技術(即人機介面)逐漸重要。在多樣人機介面當中,自然的人機介面尤為重要,在手持式計算裝置及平版電腦上的手寫或手勢是人機介面中較自然的方式,因此本論文將對手寫軌跡以及手勢辨識進行研究。由於此類的人機介面自由度較高,我們利用傅立葉描述元(Fourier descriptor)以及shape context,皆為平移、旋轉、縮放等rigid transformation下維持不變的方法。在手繪圖形,我們收集114位使用者的手繪資料,繪圖的過程中,依使用者直觀的方式,繪圖於電腦的觸控板,而這些使用者幾乎為首次使用觸控筆。當我們利用傅立葉描述元時,可達到67%辨識率;而使用shape context時,有90%的準確率。另外,我們將此技術應用於手勢辨識,收集348張手勢的照片,同樣使用傅立葉描述元以及shape context,其辨識率各為62%以及70%。
由於我們可以利用以上二方法定義出距離,即可使用K-Nearest Neighbor(KNN)為分類的方法。分別透過傅立葉描述元以及shape context所定義的距離,在辨識3D幾何物件約可達75%與95%,而在手勢辨識約有78%以及82%的辨識率。 / The cost of computing devices has dropped significantly in recent years, enabling diversified applications that require natural man-machine interaction such as pen-based computing and gesture-based communication. Whereas the automatic recognition of handwriting has been studied quite extensively, research on hand-drawn geometric shapes has received relatively little attention. In this thesis, we investigate an effective method to recognize hand-drawn geometric shapes and hand gesture. Due to the high degree of freedom of natural human-computer interface, we apply two methods, namely, Fourier descriptor (FD) and shape context (SC) to aid shape recognition. For hand-drawn shapes, we collect 114 users' free-hand drawings using Tablet PC. In this study, we achieve an accuracy of 67% by FD and 90% by SC. For gesture-based interface, we gather 348 pictures of hand gestures and obtain a classification rate of 62% by FD and 70% by SC.
Since FD and SC are distance measures, we can use K-Nearest Neighbor (KNN) classifier to improve the recognition rate. The incorporation of KNN classifier has increased the precision to 75% and 95%, where distance is measured by FD and SC respectively. For hand gestures, the improved accuracy is 78% by FD and 82% by SC.
|
3 |
從生物辨識應用探討隱私權之保護 / The privacy protection issues of biometric application游璿樺, Yu, Hsuan Hua Unknown Date (has links)
自從美國911恐怖攻擊事件後,生物辨識技術受到世界各國重視,使得生物辨識應用大鳴大放,涵蓋範圍非常廣泛,從國家的入出境管理、國民身分證,到公司或住家的門禁管理、安全監控,乃至於個人身分確認,如電腦開機登錄、隨身碟資料加密。然而生物辨識應用,會涉及個人生物特徵之蒐集與相關個人資料之運用,一方面為生活上帶來便利,另一面也無聲無息為個人隱私帶來衝擊與威脅。本文從生物辨識技術之研究,藉由分析指紋辨識、臉型辨識及DNA辨識之應用所引發的隱私權問題,以及相關法令規範之探討,最後從法制面、政策面與執行面上提供建議,希望藉由完備的法令規範,評估各種應用可能引發之隱私爭議,建立完善的管理制度與監督機制,將生物辨識應用之隱私侵害與疑慮降到最低,得以享受生物辨識應用所帶來的安全性與方便性。
|
4 |
干擾狀況下的交通標誌偵測與辨識楊修銘, Yang,Hsiu-Ming Unknown Date (has links)
在不利的環境下做交通標誌的偵測與辨識是一件非常艱困的工作,無論在郊區或市區,複雜的環境、天候、陰影以及任何和光線有關的因素甚至是交通標誌遭到遮蔽都將使得偵測與辨識交通標誌變得相當困難。在本篇論文中,我們定義出較寬鬆的顏色分類(color thresholding)方法,配合一些交通標誌的特徵(如外形)來實作出召回率(Recall)較高的偵測系統,另外在辨識方面,最重要的是找出好的辨識特徵,因此我們利用離散餘弦轉換(discrete cosine transform)和奇異值分解(singular value decomposition)處理待辨識標誌擷取其特徵,並配合一些其他的交通標誌特徵,當作類神經網路(ANN)、naïve Bayes classifier等辨識方法的輸入,來幫助我們完成辨識的工作。目前實作出來的系統在有挑戰性的測試資料下有七成六左右的辨識率。 / Robust traffic sign recognition can be a difficult task if we aim at detecting and recognizing traffic signs in images captured under unfavorable environments. Complex background, weather, shadow, and other illumination-related problems may make it difficult to detect and recognize signs in the rural as well as the urban areas. In this thesis, I define a formula for color classification and apply other related features such as the shape of the traffic signs to implement the detection component that offers high recall rate. In traffic sign recognition, the most important thing is to get the effective features. I use discrete cosine transform and singular value decomposition to collect the invariant features of traffic signs that will not be severely interfered by disturbing environments. These invariant features can be used as the input to artificial neural networks or naïve Bayes models to achieve the recognition task. This system yields satisfactory performance about 76% recognition rate when I test them with very challenging data.
|
5 |
以圖文辨識為基礎的旅遊路線規劃輔助工具 / Tour Planning Using Landmark Photo Matching and Intelligent Character Recognition黃政明, Huang, Cheng Ming Unknown Date (has links)
智慧型手機的用途已從語音溝通延伸轉變為多功能導向的的生活工具。目 前多數的智慧型手機均具備攝影鏡頭,而此模組更已被公認為基本的標準 配備。使用者透過手機,可以輕易且自然地拍攝感興趣的物體、景色或文 字等,並且建立屬於自己的影像資料庫。在眾多的手機軟體中,旅遊類的 程式是其中一種常見整合內容與多項感測模組的應用實例。在行動平台上, 設計一個影像辨識系統服務可以大幅地協助遊客們在旅途中去瞭解、認識
知名的地標、建築物、或別具意義的物體與文字等。 然而在行動平台上的可用資源是有限的,因此想要在行動平台上開發有效 率的影像辨識系統,是頗具挑戰性的任務。如何在準確率與計算成本之間 取得最佳的平衡點往往是行動平台上開發影像辨識技術的最重要課題。 根據上述的目標,本研究擬於行動平台上設計、開發行動影像搜尋與智慧 型文字辨識系統。具體而言,我們將在影像搜尋上整合兩個全域的特徵描 述子,並針對印刷與手寫字體去開發智慧型文字辨識系統。實驗結果顯示, 在行動影像搜尋與文字辨識的效能測試部分,前三名的辨識率皆可達到的 80%。 / The roles of smart phones have extended from simple voice communications to multi-purpose applications. Smart phone equipped with miniaturized image capturing modules are now considered standard. Users can easily take pictures of interested objects, scenes or texts, and build their own image database. Travel-type mobile app is one example that takes advantage of the array of sensors on the device. A mobile image search engine can bring much convenience to tourists when they want to retrieve information regarding specific landmarks, buildings, or other objects.
However, devising an effective image recognition system for smart phone is a quite challenging task due to the complexity of image search and pattern recognition algorithms. Image recognition techniques that strike a balance between accuracy and efficiency need to be developed to cope with limited resources on mobile platforms.
Toward the above goal, this thesis seeks to design effective mobile visual search and intelligent character recognition systems on mobile platforms. Specifically, we propose two global feature descriptors for efficient image search. We also develop an intelligent character recognition engine that can handle both printed and handwritten texts. Experimental results show that the accuracy reaches 80% for top-3 candidates in visual search and intelligent character recognition tasks.
|
6 |
電視新聞字幕對閱聽人處理新聞資訊的影響呂愛麗 Unknown Date (has links)
研究主要探討電視新聞字幕的數量及新聞屬性對閱聽人的新聞內容辨識及回憶程度的影響。研究結果顯示,收看低程度字幕的具體/圖像新聞的受試者的新聞內容辨識程度比收看低程度字幕的抽象/文字新聞的受試者好;這證實具體的影像或圖像確實對閱聽人辨認新聞內容有正面幫助。
受試者在收看新聞時對字幕的注意程度,以及不同程度的注意會否影響新聞內容辨識及回憶程度是本研究另一個研究目的。結果顯示,收看抽象/文字新聞,高程度字幕組的受試者,他們對跑馬燈的注意程度與新聞內容辨識程度呈現正相關,這表示越注意跑馬燈內容的受試者,新聞內容辨識程度越好;對地標的注意程度與新聞內容回憶程度則呈現負相關,顯示越不注意地標的受試者,新聞回憶程度越不好。至於低程度字幕組的受試者,他們對地標的注意程度與新聞內容辨識程度呈現正相關,這表示他們越注意地標,其新聞內容辨識程度越好。
性別是否對新聞內容辨識及回憶程度有影響也是本研究探討的一部分。研究結果顯示只有收看抽象/文字新聞,低程度字幕組的男受試者與女受試者的回憶程度有顯著差異,而且男受試者比女受試者好。
從各項研究結果的呈現足以發現,在統計上有顯著差異的都涉及抽象/文字新聞,無論是針對新聞屬性、對字幕的注意程度、或者性別,這顯示或許字幕的數量並不是影響閱聽人記憶的充分條件,新聞屬性才是關鍵。換言之,電視新聞製作人在製作字幕,或者設計鏡面上,美學不應該是唯一考量的因素,必須將新聞屬性納入考慮,配合不同的設計,對閱聽人學習電視新聞才能達到正面幫助。
|
7 |
中文繁簡等義詞自動辨識之研究 / A Study on Automatic Recognition on Exact Synonyms between Traditional and Simplified Chinese黃群弼 Unknown Date (has links)
中文繁簡在字體或電腦編碼上明顯不同之外,在部份詞彙的用法也有不同,這些用法不同的詞彙卻有相同意義的詞彙稱為繁簡體中的等義詞,這些等義詞在雙方文化交流時可能會造成一些障礙,例如人們互相對話、文件書籍翻譯或軟體系統等轉換時容易造成詞義上的誤解,目前均以人工方式來解決不同詞彙的問題,均會費時耗力且易疏漏,若能利用科學的方法讓電腦能自動辨識中文繁簡的等義詞,便能利用辨識出的等義詞給予提示,解決繁簡詞義不同所造成的誤解。
依照實驗設計架構,首先建立電腦類與一般類的繁簡體語料庫,作為辨識的基礎,並建立研究的架構與方法,分為二個階段三種方法,第一階段使用第一種方法,我們先使用N-gram辨識等義詞,評估單一方法是否能有效辨識出等義詞,第二階段使用第二種方法PMI-IR & LC-IR方法與第三種方法Context Vector,評估第二階段的方法是否能將等義詞的辨識能力提高。
根據本研究目的,讓電腦能自動在語料庫中自動辨識中文繁簡等義詞,所以提出了新的辨識架構,用N-gram初步辨識出等義詞,並經由PMI-IR & LC-IR與Context Vector方法提高Precision約0~20%不等。本研究結論是採用不同語言的語料庫,使用N-gram能夠辦識出等義詞,並搭配PMI-IR & LC-IR與Context Vector方法後,可以強化與提昇其等義詞辨識的能力,解決單一方法等義詞辨識能力不足之問題。 / Traditional Chinese and Simplied Chinese are not only different in the typeface and in the computer code, but also in the partial usage of vocabularies. These vocabularies which have different usage but have the same significance are called synonyms. These synonyms will cause some obstacles and misunderstanding in meaning when two parties have cultural exchange, such as during conversation, documents and books translation or softwares system transformation. What we do to solve the problem now is picked them out by manpower, but that will waste a lot of time and strength and easily make errors. If we can use scientific way to let the computer distinguish automatically the synonyms between Traditional Chinese and Simplied Chinese, we will be able to solve such misunderstanding by the hints of the distinguished synonyms.
According to the structure of experiment, to let the computer distinguish automatically the synonyms between Traditional Chinese and Simplied Chinese, we have to establish a Traditional Chinese and Simplied Chinese computer category and a general category first as the basis of identification. We should build up the research structure and the method, which divided into two stages and three methods. The first stage uses the first method to use N-gram to distinguish the synonyms and then review if this single method can identify the synonyms effectively. The second stage uses the second method PMI-IR & LC-IR and the third method Context Vector and review if the second stage can raise the synonyms’ ability of identification.
According to this research purpose, the computer to study on automatic exact recognition synonyms between traditional and simplified Chinese, so has proposed the new structure of distinguishing, N-gram automatic exact recognition synonym tentatively, and PMI-IR & LC-IR and Context Vector method can improve Precision about 0~20%. This conclusion is a corpus base of using different languages, using N-gram can be exact recognition synonyms, PMI-IR & LC-IR and Context Vector method, can improve single method ability.
|
8 |
無限射頻辨識系統(RFID)導入的成功關鍵因素探討 / Key Success Factors Analysis in the Implementation of RFID Technology劉俊良, Liu, Eric Unknown Date (has links)
現在越來越多的廠商和美國國防部都需要較好的無限射頻辨識系統(RFID)系統,所以它越來越受到注目,本論文即在探討無限射頻辨識系統(RFID)導入成功的關鍵因素。 / With the driving force from Wal-Mart, the world’s largest retailer and the US Department of Defense, suppliers are required to integrate Radio Frequency Identification (RFID) in their case and pallet shipments to distribution centers in their supply chain. In comparison with traditional bar code labels and magnetic strips for supply chain management, RFID technology offers better visibility and information integration in the supply chain management.
In this paper, a variety of Automatic identification technologies will be compared to demonstrate the advantages of RFID solutions. The introduction of RFID technology will be made as well to detail the components in the RFID systems and the factors taken into account in the RFID selection and design phases. In the technology implementation, the diffusion model is adopted to explain the evolution of new technology implementation process. The strategic model for the adoption of RFID technology in business process and management is presented as a guideline for companies who are considering adopting the RFID solutions. The impact on business management and the practice guideline to the RIFD implementation are illustrated.
The factor analysis in the driving forces and resistance to the RFID adoption are examined to identify the attributes of its successful implementation and the variables of its adoption. According to the factor analysis, the driving forces are summarized into four major ones, including technology innovation, government and standard organization influence, organizational readiness and inter-organization demands. Two cases of RFID applications are presented to illustrate the factors taken into account in the RFID implementation. These two cases include the health care application and agricultural product process application. Companies gain the benefits of the improvement in the production efficiency and quality control over the business process and management. Information flow and capturing are becoming visible and automatic with the implementation of RFID technology.
|
9 |
自閉症兒童臉孔情緒處理之研究蔡佳津 Unknown Date (has links)
臨床上我們觀察到:自閉症患者在社會互動上有明顯障礙。因此,我們企圖在這個大問題下,以社會互動中最主要的訊息來源—臉部表情為出發點,探討自閉症者所知覺的世界中,他們如何處理臉部表情所展現的情緒訊息及社會意義。
根據文獻回顧,我們知道:臉孔辨識歷程與一般物體辨識歷程不盡相同而有其特性。相較於一般物體,臉孔辨識更需依賴對輪廓構型訊息的有效掌握。因此,本研究企圖回答以下問題:「自閉症患者在辨識他人臉部表情上是否有所缺陷」?如果是的話,(1)「自閉症患者在辨識他人臉部表情上的缺陷係因臉孔辨識的機制上有所缺陷嗎」?還是(2)「自閉症患者在辨識他人臉部表情上的缺陷是因從他人臉部表情讀取情緒訊息上有所困難」?本研究試圖以人臉辨識與表情辨識的差異性,並改良過去研究在方法學上的爭議以回答這兩個問題。
實驗一「人臉辨識」作業以立即比對作業與延宕配對作業,檢驗自閉症組、發展遲緩組、一般兒童組及成人組其在物體與臉孔之倒立效果。研究結果顯示:除了自閉症組外,三組受試者在人臉辨識的作業表現上,都有相當穩定的臉孔倒立效果,自閉症組的臉孔倒立效果顯著地較其它三組小,而在物體辨識上與常人無異,此與「中樞連貫缺陷」假設相符應:自閉症患者在高層次知覺的困難在於他們無法從環境中將訊息加以整合,擅於以局部特徵來理解,並認為這是使他們產生較小的臉孔倒立效果之因。實驗一顯示:自閉症患者的確「在辨識他人臉部表情上有所困難」,卻「並非因其在臉孔辨識的機制上有所缺陷」。
因此,我們第二個研究問題即是:「自閉症患者在辨識他人臉部表情上的缺陷是因從他人臉部表情讀取情緒訊息上有所困難嗎」?因此,實驗二以同一組臉孔刺激材料,以人臉辨識與情緒辨識作業,以探討自閉症組、一般兒童組與發展遲緩組的作業表現。從實驗二所得到的結果也支持這樣的看法。在實驗二的「人臉辨識」作業中,要求受試者對目標人臉進行辨識時,三組受試者所受到的「臉部表情干擾效果」並無顯著差異。但在實驗二的「表情辨識」作業中,當要求受試者對目標情緒進行辨識時便發現:自閉症組在辨識臉部表情上的缺陷,極大部分是發生在辨識不同人的表情變化情境下。而之所以自閉症患者在辨識同一人的表情相同與否表現較佳,依據「中樞連貫缺陷」假設的看法則認為,自閉症患者使用他們擅長以局部特徵瞭解整體的能力克服在他們在處理作業上的困難。此外,實驗二對兩作業表現的比較結果亦支持:大腦對於辨識人臉以及辨識情緒顯示是由不同的系統進行處理。
因此從本研究的兩個實驗,我們可以清楚地回答:自閉症患者在辨識他人臉部表情上的困難確實非因臉孔辨識的機制上有所缺陷而是從他人臉部表情讀取並理解情緒訊息上有所困難,尤其是發生在辨識不同人的表情變化情境下。而之所以自閉症患者在辨識同一人的表情表現較佳,是因為自閉症患者使用其擅長以局部特徵瞭解整體的能力克服在他們在處理作業上的困難。而自閉症患者在這些作業中的種種行為表現,都可以運用「中樞連貫缺陷」假設得到不錯的解釋。
|
10 |
植基於質感圖樣之自動化人機區分機制 / A CAPTCHA Mechanism Based on Textured Patterns張繼志, Chi-Chih Chang Unknown Date (has links)
隨著科技的進步與資訊科學的發展,大量的資訊處理自動化逐漸取代傳統人工技術,然而不恰當地使用自動化技術,卻可能危害人類的權益與空間。為避免過度濫用機器自動化對人類所造成的災害,本研究根據不同的適用情境,分別提出以靜態及動態圖型為基礎的人機區分方法,透過簡單的影像處理技術,產生機器難以分析但人類能夠易於判別的人機辨識影像。並且由認知的角度,設計實驗進一步探討人類視覺優勢以及接受度,作為影像產生時的標準。最後,提出人機區分技術與應用情境整合實作的方法,以觀實效。 / The idea of using a computer program to distinguish humans from machines, sometimes referred to as the “Reverse Turing Test”, has emerged only quite recently. The term CAPTCHA, which stands for “Completely Automated Public Turing Test to Tell Computers and Humans Apart", is defined as:
“a program that can generate and grade tests that:
□ Most human can pass
but
□ Current computer program can’t pass! “
In this thesis, a texture-image based approach is developed to encode text information in such a way that machine vision algorithms will experience significant difficulties while human can extract the embedded text effortlessly. Both static images and dynamic sequences will be explored. It is anticipated that the cost of storing, and subsequently decoding information from such visual patterns will be prohibitedly high, both in terms of time and space complexity. To validate the postulation, fundamental principles of the human cognitive process will be examined. Experiments will also be carried out to gather user feedback and investigate the limitations of human visual systems. Finally, several application scenarios that call for the integration of a CAPTCHA will be identified and discussed.
|
Page generated in 0.0283 seconds