一种基于视觉知识加工模型的目标识别方法

发布时间：2018-07-12 18:53

本文选题：目标识别 + 视觉知识　；参考：《自动化学报》2016年05期

【摘要】：提出了一种基于视觉知识加工模型的目标识别方法.该加工模型结合目标定位、模板筛选和MFF-HMAX(Hierarchical model and X based on multi-feature fusion)方法对图像进行学习,形成相应的视觉知识库,并用于指导目标的识别.首先,利用Itti模型获取图像的显著区,结合视觉通路中What和Where通道的位置、大小等特征以及视觉知识库中的定位知识确定初期候选目标区域;然后,采用二步去噪处理获取候选目标区域,利用MFF-HMAX模型提取目标区域的颜色、亮度、纹理、轮廓、大小等知识特征,并采用特征融合思想将各项特征融合供目标识别;最后,与单一特征以及目前的流行方法进行对比实验,结果表明本文方法不仅具备较高的识别效果,同时能够模仿人脑学习视觉知识的过程形成视觉知识库.
[Abstract]:A method of target recognition based on visual knowledge processing model is proposed. The processing model combines target location, template screening and MFF-HMAX (hierarchical model and X based on multi-feature fusion) method to study the image, and form the corresponding visual knowledge base, which can be used to guide target recognition. Firstly, the salient region of the image is obtained by using the Itti model, the location and size of the what and where channels in the visual path are combined with the location knowledge in the visual knowledge base to determine the initial candidate target areas. Using two-step denoising process to obtain candidate target region, MFF-HMAX model is used to extract knowledge features of target region, such as color, brightness, texture, contour, size and so on. Finally, features are fused for target recognition by feature fusion idea. Compared with the single feature and current popular methods, the results show that this method not only has a high recognition effect, but also can imitate the process of learning visual knowledge in human brain to form a visual knowledge base.
【作者单位】：上海海事大学信息学院;
【基金】：国家自然科学基金(31170952) 国家海洋局项目(201305026) 上海海事大学优秀博士学位论文培育项目(2014bxlp005);上海海事大学研究生创新基金项目(2014ycx047)资助~~
【分类号】：TP391.41

【参考文献】