对象定位处理中分类信息融合技术研究

Research of fusing image classification into object localization

为提高图像中对象定位技术的处理效果,对对象定位技术和分类技术的融合进行了研究。针对大规模、多对象类别的图像对象定位问题,提出了先进行快速分类,再精确定位的处理方案。通过MIMLSVM+多类别分类算法预判出包含对象的图像,利用ESS方法在上述图像中定位对象;针对高精度对象定位需求,提出了融入全局分类信息的最优框打分机制,将MIMLSVM+算法对于图像的分类信息融入ESS方法中最优框的打分信息中。在PASCAL 2006数据集上相应的实验结果表明,前者在缩短处理时间的同时取得了不错的定位平均精度,而后者对最优框得分的改进也在多个类别上带来了定位效果的提高。实验结果表明,分类信息融入对象定位处理中能提升处理效果。

In order to improve the performance of the object localization technology, this paper researched the approach of fusing image classification into object localization. According to the object localization task in the large-scaled image set containing multi-class objects, this paper proposed an effective scheme that a fast image classification was carried out before a precise object localization. The MIMLSVM ＋ algorithm predicted which images contained the objects, then the ESS method localized the objects just in those images. And aiming at a higher precision in the object localization task, the paper proposed a new scoring mechanism of the optimal box which included the global classified information in the multi-label classification task. The corresponding experimental results on the PASCAL 2006 data set show that the former method shortens the processing time while obtaing a good average positioning accuracy; the latter method brings an improvement in the localization performance on many categories as it make the scoring mechanism of the optimal box better. The experimental results prove that fusing image classification into object localization can really improve the performance of object localization.