Abstract: For the problem that the salient map based on color histogram can’t highlight the edge contour and texture details, a method of image saliency detection based on SLIC fusion of texture and histogram is proposed in this paper, which combines color feature, spatial location feature, texture feature and histogram. Firstly, the SLIC algorithm is used to segment the image and the saliency map based on color and spatial location is extracted. Then, saliency map based on color histogram and saliency map based on texture feature are extracted respectively. Finally, the saliency maps obtained from the first two stages are fused to get the final saliency map. In addition, the saliency target in the image is obtained by a simple threshold segmentation method. The experimental results show that compared with the classical saliency detection algorithms, the performance of the algorithm is much better than that of other algorithms.

Key words: SLIC algorithm, color feature, spatial location feature, texture feature, histogram, saliency detection

摘要： 针对基于颜色直方图的显著图无法突出边缘轮廓和纹理细节的问题，结合图像的颜色特征、空间位置特征、纹理特征以及直方图，提出了一种基于SLIC融合纹理和直方图的图像显著性检测方法。该方法首先通过SLIC算法对图像进行超像素分割，提取基于颜色和空间位置的显著图；然后分别提取基于颜色直方图的显著图和基于纹理特征的显著图；最后将前两个阶段得到的显著图进行融合得到最终的显著图。此外，通过简单的阈值分割方法得到图像中的显著性目标。实验结果表明，与经典显著性检测算法相比，提出的算法性能明显优于其他算法性能。

关键词: SLIC算法, 颜色特征, 空间位置特征, 纹理特征, 直方图, 显著性检测