Current pornographic images filtering algorithms have some shortcomings, such as high false positive rate toward the bikinis images and insufficiency when filtering pornographic images with pornographic actions. The paper proposed a new pornographic image filtering model based on High-level Semantic Bag-of-Visual-Words (BoVW). Firstly, local feature points in sex scene were detected using the Speeded-Up Robust Features (SURF) algorithm and then high-level semantic dictionary was constructed by fusing the context of the visual vocabularies and spatial-related high-level semantic features of pornographic images. The experimental results show that the model has an accuracy up to 87.6% when testing the multiperson pornographic images, which is significantly higher than the existing pornographic images filtering algorithm based on BoVW.%针对目前色情图像过滤算法对比基尼图像和类肤色图像误检率过高,且不能有效过滤带有淫秽动作的多人色情图像的缺点,提出一种基于高层语义视觉词袋的色情图像过滤模型.该模型首先通过改进的SURF算法提取色情场景局部特征点,然后融合视觉单词的上下文和空间相关高层语义特征,从而构建色情图像的高层语义词典.实验结果表明,该模型检测带有淫秽动作的多人色情图像准确率可达87.6%,明显高于现有的视觉词袋色情图像过滤算法.
展开▼