資訊科技的發展方向漸漸由企業導向轉為居家生活導向。以往的居家安全大多是顧用保全、警衛或是架設數位監控系統對居家環境進行監控,當發現疑似入侵者時,系統則進行錄影動作或發出警報。但是此種方法的缺點就是當保全人員疲累或疏忽時,易造成安全上的漏洞,且目前大部分的監控系統只擁有偵測移動物件之功能,並無自動判斷出入者是否為家庭成員的功能,因此若有一套可即時偵測並自動判斷出入者是否為家庭成員的監控系統,如此將可避免上述之問題。本論文提出一個結合影像識別與聲音識別技術的自動化即時門禁監控系統,利用每個人與生俱來的生物特徵資訊進行家庭成員的辨識,並且能即時判斷進門者是否為入侵者。
Successfully recognizing family or organization members is important for security. Traditionally, the recognition tasks are the duties of the guards by directly surveilling the incoming people or monitoring the video. Due to the fast progress of information techniques, the speaker, face and other biometric characteristics recognition have achieved great success. In this paper, we introduce a real-time family member recognition system using based on images and voices. These images and voices are fetched by low price cameras and microphones. This surveillance system will automatically detect the intruders and emit the alarms. Testing with five family members and ten intruders in the field, our system can achieve 88.3% recognition.