多層式解析分割之影像切割之研究

Abstract

本篇論文主要在研究的方向是將一已存在的影像切割的演算法Grab cut，使用一種階層式的方法來做預先運算，可使得原來的方法在疊代的過程中，收斂的更快。原來的Grab cut使用資料分群的概念做為切割的原理，在疊代的過程中，有時會疊代非常多次才達成收斂，導致整個運算必須花很長的時間，所以我們使用了預處理的概念，首先用低解析度的影像來做疊代到收斂，如此可以大幅降低運算量，而停止時的切割結果，由於顏色組成不會有大的變化，將近似原解析度影像疊代到停止時的結果，所以我們便將此時的分群後的數值交給原來解析度的影像做接續的疊代，便能用較少的疊代次數完成此程序。此外，我們還比較了4種降低解析度的方式，最後發現，先用Median Filter處理降低Watershed的集水盆數的階層式影像切割法，有最好的速度表現。

This thesis aims to speed up a segmentation algorithm -- Grab Cut -- by separating the processing of segmentation into hierarchical steps. The Grab Cut algorithm segments images with the color clustering concept which needs a lot of times of iteration for it to converge. It is a time consuming processing that we are interesting in the improvement of this processing. Therefore, we utilize the idea of hierarchical processing, that is, the first step is to compute at low resolution to make the iteration much faster, and the second step is conducted based on the result of the first step to carry on iteration at original resolution so that the total execution time can be reduced. More detailed, Segmentation of a low resolution image will result in high-speed and similar-segmentation result to the segmentation at original resolution. Hence, after the iterations at low resolution has converged, we utilize the parameters of segmentation result to initialize the next segmentation on original resolution. This way, the number of iteration of segmentation at original resolution will be reduced through the initialization of those parameters. Since the execution time with low resolution images is relatively short, the total hierarchical execution time will be reduced consequently. Besides, the author made a comparison between the four methods of reduction on image resolution. Finally, we found that reducing the number of basins by “Median Filter” resulted in best segmentation speed.