The algorithm acquires the quantity of two-dimensional spatial translation by means of projecting each frame apart along horizontal and vertical direction and using multi-resolution spectrum decomposition approach to process the two projection vectors, respectively 該算法首先引入積分投影方法將圖像序列中的每幀圖像轉(zhuǎn)換成兩個一維積分投影向量,再應(yīng)用多分辨譜分解方法對兩個投影向量分別進行處理,從而精確地獲取圖像在二維空間上的亞像素平移量。