基于局部直方图信息的相似图像组分割模型
The model for segmenting group of similar images based on local histogram information

针对ACGS模型不能有效分割具有纹理变化的相似图像组问题，提出一种基于图像局部直方图信息的协同分割模型. 该模型能量泛函分为两项，一项为基于图像直方图信息的数据项，反映了图像组中每张图像的纹理信息；另一项为用形状矩阵的秩表示的相似性约束项，用来控制待分割图像之间目标形状的相似性，其中形状矩阵中每个列向量代表图像组中一幅图像的目标轮廓. 同时，在SUN显著图上演化CV模型实现轮廓的初始化，提高了ACGS模型对初始轮廓位置的鲁棒性. 实验表明，该模型对具有纹理变化的形状相似图像组分割效果优于ACGS模型.
Aiming at solving the problem that ACGS model cannot effectively segment the group of similar images with texture changes，this paper proposes a co-segmentation model based on image local histogram information. The energy function of the proposed model consists of two items，one is the item based on image local histogram information，reflecting the texture information of each image in group of similar images；the other one is represented by the rank of the matrix consisting of multiple shapes，used to control the similarity of object shapes. And each column of the shape matrix represents an image in the image group. In addition，we get the initial contours of the proposed model by using CV model on images SUN saliency maps，which increases the robustness of the ACGS model against initial contours. The experimental results show that the proposed model has a better performance on the segmenting result than the ACGS model，for the group of similar images with textured change