基于字典优化的稀疏表示的视频镜头分类

Video shot classification based on sparse representation of dictionary optimized

为了克服稀疏表示中冗余字典分类效果不佳的问题,提出了基于字典优化的稀疏表示算法。该算法制定了新的基于稀疏表示的分类判别规则,采用了基于冗余字典内基元类内平均欧式距离最小以及类间平均欧式距离最大的字典优化方法,形成优化字典进行特征稀疏表示。将该算法应用于视频镜头的稀疏表示特征提取与分类,实验结果表明该方法优化后的字典进行视频镜头的特征提取和分类,其识别率得到了明显的提高。

In order to overcome the ineffective classification results of the redundant dictionary in the sparse representation-based classifier,this paper presented a sparse representation algorithm based on dictionary optimization.The algorithm developed a new classification discriminate rules based on sparse representation.It optimized the dictionary by the method of minimizing the average of the in-class Euclidean distance and maximized the average of the between-class Euclidean distance,formed the optimized dictionary and presented the features based on sparse representation.And the algorithm was applied on video shot to extract feature and classify based on sparse representation.The experimental results show that the recognition rate of feature extraction and classification on video shot based on the dictionary optimized by this method has been significantly improved.