在即時之MPEG-4監視系統上植基於內容的視覺物件編碼之運用決定機制

Abstract

在影像處理技術的急速進步之下，以物件為基礎的影像編碼技術也漸漸廣泛地使用於交通監視系統。然而，以物件為基礎的編碼方法在遇到切割出來的物件較大時，也得花費較多的計算量使得編碼速度緩慢。這篇論文展示了一個套用在視覺基礎智慧型運輸系統上的編碼方法切換架構，以這個架構來決定當下是要使用以物件為基礎的編碼方法或者是以視幀(frame)為基礎的編碼方式。而實驗結果也顯示出這個切換架構的確對系統的運作效率有所幫助。

With the rapid development of video technology , traffic surveillance systems using object-based encoding techniques become popular. However, object-based encoding may takes large computation power while segmented object is large. That situation causes performance of system slow down. This paper presents an encoding method switching scheme on a vision-based intelligent transportation system in order to determine when to encode by frame-based encoding method or to encode by object-based encoding method. The experimental results show that the switching scheme on traffic surveillance system is indeed effective for system performance.