中文摘要：

为了选择有效的图像特征,并将这些特征融合以进行图像的显著区域检测,提出一种基于图像特征稀疏约束的显著性检测算法。该算法首先建立一个包括多种图像特征的特征池,之后假设图像的显著图由特征池中特征的线性组合来表示,并用线性回归的方法从眼动追踪数据库的信息中学习出该线性组合的权重参数;在学习的过程中,对线性回归的系数加一个稀疏约束条件,使得某些不重要特征对应的系数在最优化过程中自动收缩为0,从而达到特征选择的目的。实验结果表明,该模型的检测时间较短,可以得到较高的检测准确率。与传统基于特征融合的显著性检测模型相比,本算法避免了选择特征和构造融合参数的盲目性。

英文摘要：

In order to select effective image features and integrate them to detect the salient region of an image,this paper proposed a saliency detection algorithm based on sparse constraint of image features. The algorithm firstly established a feature pool which included a variety of image features,then supposed that image＇s saliency map can be represented by a linear combination of features in the feature pool and learned the weight parameters of this linear combination from the information of eye tracking database; during the process of learning,adding a sparsity constraint to the coefficients of linear regression,making some coefficients corresponding to the unimportant features shrink to 0,so as to achieve the purpose of feature selection. The results of experiments show that the detection time of this model is short and the detection accuracy is high. Compared with traditional saliency detection model based on feature integration,the algorithm avoids the blindness in feature selection and integration parameter construction.