Chuẩn bị dữ liệu trong việc xây dựng cơ sở dữ liệu video số

In this paper, a method to prepare video data is proposed consisting of a shot boundary detection technique on the basis of tracking camera motion through the background areas to the video frame (background tracking) and of some other steps to discard noise and to reduce false segmentation for increasing the accuracy of the video stream segmentation. Measures to be taken to overcome the problem of object’s and camera’s motion are proposed and an enhanced algorithm for segmenting video data is also reported. | ’ Tap ch´ Tin hoc v` Diˆu khiˆn hoc, , (2007), 141–152 ı e e . . a ` . . ˆ . ’ ˆ ˆ ˆ CHUAN BI DU LIEU TRONG VIEC XAY DU NG . . . . ˜ . ’ . ˜. ˆ ´ ˆ CO SO DU LIEU VIDEO SO . . . . . ˜ ˜ ´ ´ ˆ ˆ ˆ ˆ ˆ ˆ LU O NG XUAN CU O NG1 , DO TRUNG TUAN2 , DO XUAN TIEN1 1 Hoc viˆn K˜ thuˆt Quˆn su., Email: lxcuong2002@; dxtienvxl@ e y a a . . . . 2 Dai hoc Quˆ c gia H` Nˆi, Email tuandt@ ´ o a o . . . Abstract. Preparing video data is an important step in the construction of digital video databases. This work demands segmenting a video sequence into shots, which are basic meaning units of video data. In this paper, a method to prepare video data is proposed consisting of a shot boundary detection technique on the basis of tracking camera motion through the background areas to the video frame (background tracking) and of some other steps to discard noise and to reduce false segmentation for increasing the accuracy of the video stream segmentation. Measures to be taken to overcome the problem of object’s and camera’s motion are proposed and an enhanced algorithm for segmenting video data is also reported. ´ ’ ’ u e T´m t˘t. Chuˆ n bi d˜. liˆu video l` mˆt bu.´.c quan trong trong qu´ tr` xˆy co. so. d˜. liˆu o a a . u e a o o a ınh a . . . . . ˜ ´ o o e ` ’ o (CSDL) video sˆ . N´ luˆn yˆu cˆu phˆn chia chuˆi video th`nh c´c lia (shot), l` vi co. ban c´ o a a o a a a . ’ . u e ` ´ e a o ngh˜ cua d˜. liˆu video. B`i b´o dˆ xuˆt mˆt ph´p chuˆ n bi d˜. liˆu video, bao gˆ m: k˜ ıa ’ u e a a ` a a o y . . . .i gi˜.a c´c lia video trˆn d´nh dˆu v` ng nˆn cua c´c khung video, mˆt ´ ` ’ a e a thuˆt ph´t hiˆn gianh gi´ u a a a e o a u e o . . . . ˜ a a ` ´ . ’ ’ a sˆ biˆn ph´p giam nhiˆu v` c´c ph´t hiˆn sai nh˘ m t˘ng kha n˘ng phˆn chia mˆt c´ch ch´ x´c o e a e a e a a a o a ınh a . . ’ o ´ ´ e ´ ’ a a d`ng video; mˆt sˆ biˆn ph´p kh˘c phuc vˆ n dˆ chuyˆn dˆng cua camera v` dˆ i ; thuˆt to´n o o o e a a e a o . . . ´ . .