【摘要】 提出一种分割灰度不均匀单目标相似图像组的新模型。该模型结合组相似活动轮廓模型(ACGS)和不均匀性嵌入式活动轮廓模型(InH_ACM),同时考虑了图像的灰度、纹理和形状的特征,能有效分割灰度不均匀的图像组,处理图像中的误导信息。另外,采用在GB显著图上自动定位初始轮廓的方法,重新定义了平衡灰度一致性能量和纹理一致性能量的权重。实验结果表明,该方法在分割精度和收敛速度上都有优势。更多还原

【Abstract】 In this paper, we propose a novel model to segment group of similar images with intensity inhomogeneity and single object. The model is the combination of active contours with group similarity(ACGS) and inhomogeneity-embedded active contour(InH_ACM), and considers intensity, texture and shape features simultaneously. The proposed model can segment group of images with intensity inhomogeneity effectively, and cope with misleading features in images. Moreover, we use the method to automatically locate the initial contour on GB saliency map, and redefine the weights used to balance the intensity coherence energy and texture consistency energy. Experimental results indicate that proposed method is superior in segmentation accuracy and convergence speed.更多还原