CAJ | 학술논문

研究 GrabCut 是以迭代能量优化算法为基础，以颜色和纹理为特征，从背景图像中提取出目标的图像分割算法。但该方法速度较慢，为了达到实时应用的目的，提出了一种基于小波变换的 GrabCut 图像分割方法。该方法首先对原始图像进行小波变换的图像压缩，使得原始图像的分辨率降低，然后在压缩后的图像上迭代GrabCut算法，最后将收敛获得的目标区域作为原始图像初始值，再次使用GrabCut算法在原始图像上进行迭代，从而提取出目标。实验结果证明该算法提高了图像目标的提取速度，并保持图像目标的特征基本不变。
연구 GrabCut 시이질대능량우화산법위기출，이안색화문리위특정，종배경도상중제취출목표적도상분할산법。단해방법속도교만，위료체도실시응용적목적，제출료일충기우소파변환적 GrabCut 도상분할방법。해방법수선대원시도상진행소파변환적도상압축，사득원시도상적분변솔강저，연후재압축후적도상상질대GrabCut산법，최후장수렴획득적목표구역작위원시도상초시치，재차사용GrabCut산법재원시도상상진행질대，종이제취출목표。실험결과증명해산법제고료도상목표적제취속도，병보지도상목표적특정기본불변。
GrabCut is an algorithm based on the energy optimization algorithm. It uses color and texture features to extract the target from the background image. However, the processing speed of this method is slower than we expect. In order to accelerate the image segmentation, this paper presents a GrabCut image segmentation method based on wavelet transform. The proposed method compress the original image based on wavelet transform to reduce the image resolution and applies the GrabCut algorithm on it. Then identify the initial value of the original image with the convergent target area, and the algorithm applies the last GrabCut iteration on original image. Experimental results show that the algorithm improves the extraction rate of the target image, and keeps the basic characteristics of the target image unchanged.