本研究為發展透過比對相似度找尋相似背景之演算法，以快速而準確地偵測出複雜背景下之相似背景。本研究一開始先取得欲偵測背景之數位影像，接著利用影像中的位置資訊進行影像分割，將一大範圍背景影像切割為人形大小左右的小範圍影像，然後透過各小範圍影像特徵值像素比對，找出欲想要融合度範圍內最大的相似度背景位置，並進行整張背景所有相似融合度位置之標定。本研究利用此自行開發之演算法，應用於複雜背景之數位影像上，證明能快速且正確地偵測出相似融合度的區域。

The study presents an algorithm for auto-detecting similar similarity background in complex backgrounds. This algorithm includes certain techniques such as similarity calculation and image separation technology. First, we take the background images, and divide the images into many human shape size regions. Then a similarity calculation is used to find out the regions that have higher similarity. Finally, all areas of background are calculated to find the region we demand. The results of several background images show that the proposed algorithm works efficiently and accurately for the same similarity regions in complex backgrounds.