Foreground detection is an important step in various video content analysis systems such as object tracking, recognition and counting. Due to the limitations of each algorithm based on its merits and demerits, so far, there is no consensus on the most effective method due to varying nature of videos. Accuracy and timely computational processing are the two main constraints. Whilst other methods only detect the approximate motion part(s) of object(s) in a video, this paper presents a novel approach to detect the motion part(s) and associated object(s) to get the whole subject. Our work detects foreground by using a new automatic masking technique. The proposed technique uses a set of morphological operators to separate foreground and background. The proposed algorithm is an extension of previous works [1-3]. A complex video sequence was tested to detect comprehensive foreground regions of moving object(s).