【摘要】：隨著大數(shù)據、傳感器和深度學習等技術的發(fā)展,人工智能將成為各行各業(yè)的顛覆性力量。人工智能在某些特定的任務,如計算機視覺和語音識別等領域中取得了優(yōu)異的成果。但是在面對復雜場景時,人工智能仍然容易被復雜的場景或物體迷惑,盡管這些場景對于人類來說非常簡單。因此,人工智能在應用于日常生活的過程中仍然面臨自主性差、人機交互不良等問題�；谌斯ぶ悄芩媾R的自主性差、人機交互不良等問題,本文中以"機器自主感知人體,環(huán)境自主適應人體"為核心,提出了"智能感知"的概念。通過對"智能感知"的基本概念、組成和應用三方面進行詳細的分析,闡述了如何將"智能感知"應用于各行業(yè)�；�"智能感知",通過人與機器和環(huán)境三者的交互感知、交互理解,實現(xiàn)在無人為輸入信息情況下的隨時隨地、不知不覺,機器主動感知人體、環(huán)境自主適應人體。以"智能感知"概念的重要組成部分——人體狀態(tài)感知為目標,分別對如何通過攝像頭代替專用傳感器實現(xiàn)人體狀態(tài)信息的獲取,以及通過算法從數(shù)據中挖掘出表征人體狀態(tài)的特征量進行了研究。通過利用攝像頭代替血壓和心率的專用傳感器,實現(xiàn)了基于高速攝像機的動態(tài)血壓的非接觸獲取,以及基于手機攝像頭的心率的非接觸獲取。采用深度神經網絡實現(xiàn)了簡單的說話人識別,并通過非線性方法對新生兒哭聲中所包含的生物特征進行挖掘,得到了可以用于表征個體的情緒狀態(tài)和神經系統(tǒng)活動的復雜度的特征量。
[Abstract]:With the development of big data, sensor and depth learning technology, artificial intelligence will become a disruptive force in various industries. Artificial intelligence has achieved excellent results in certain tasks, such as computer vision and speech recognition. But in the face of complex scenarios, artificial intelligence is still prone to be confused by complex scenes or objects, although these scenarios are very simple for humans. Therefore, artificial intelligence still faces the problems of poor autonomy and poor human-computer interaction in the process of application in daily life. Based on the problems of poor autonomy and poor human-computer interaction in artificial intelligence, this paper puts forward the concept of "intelligent perception" with the core of "machine autonomous perception of human body and environment autonomous adaptation to human body". Through the detailed analysis of the basic concept, composition and application of "intelligent perception", this paper expounds how to apply "intelligent perception" to various industries. Based on "intelligent perception", through the interactive perception and mutual understanding between human, machine and environment, the machine can realize the human body automatically and adapts to the human body in the situation of no one input information at any time, any place, unconsciously, the machine actively perceives the human body, and the environment adapts to the human body autonomously. Aiming at the important component of the concept of "intelligent perception", human state perception, how to obtain human state information by using camera instead of special sensor is discussed respectively. And the algorithm is used to extract the characteristic quantity from the data to represent the state of the human body. By replacing the special sensor of blood pressure and heart rate with camera, the non-contact acquisition of ambulatory blood pressure based on high-speed camera and the non-contact acquisition of heart rate based on mobile phone camera are realized. The deep neural network is used to realize simple speaker recognition, and the biometric features contained in the crying of newborn are mined by nonlinear method. The characteristic quantities which can be used to represent the emotional state and the complexity of nervous system activity are obtained.
【學位授予單位】：浙江大學
【學位級別】：碩士
【學位授予年份】：2017
【分類號】：TP391.41;TP18

【參考文獻】

相關期刊論文 前8條

1 李申龍;李毅彬;李洪陽;張洋;陳曉萌;鄧寧;;基于脈搏波相位差的無創(chuàng)連續(xù)血壓測量方法[J];傳感器與微系統(tǒng);2016年01期

2 王繼光;;24h動態(tài)血壓監(jiān)測:適用人群及臨床意義[J];中華高血壓雜志;2014年07期

3 鄭理華;竇建洪;何興華;屠偉峰;馬銘俊;吳昊;;無創(chuàng)血壓測量技術的改進與進展[J];中國醫(yī)學裝備;2013年03期

4 李章俊;王成;朱浩;金凡;馬俊領;;基于光電容積脈搏波描記法的無創(chuàng)連續(xù)血壓測量[J];中國生物醫(yī)學工程學報;2012年04期

5 張維忠;;血壓變異和晨峰的概念及其臨床意義[J];中華心血管病雜志;2006年03期

6 向海燕;俞夢孫;;用脈搏波傳導時間實現(xiàn)血壓連續(xù)測量[J];醫(yī)療衛(wèi)生裝備;2006年02期

7 肖曉;;血壓為何會升高[J];藥物與人;2004年01期

8 張政波,吳太虎;無創(chuàng)血壓測量技術與進展[J];中國醫(yī)療器械雜志;2003年03期

，

本文編號：2204415