SYSTEM AND METHOD FOR FINDING SADDLE POINT-LIKE STRUCTURES IN AN IMAGE AND DETERMINING INFORMATION FROM THE SAME
    23.
    发明申请
    SYSTEM AND METHOD FOR FINDING SADDLE POINT-LIKE STRUCTURES IN AN IMAGE AND DETERMINING INFORMATION FROM THE SAME 有权
    用于在图像中发现类似点状结构的系统和方法以及从其确定信息

    公开(公告)号:US20140119665A1

    公开(公告)日:2014-05-01

    申请号:US13797910

    申请日:2013-03-12

    Abstract: This invention provides a system and method for finding features in images that exhibit saddle point-like structures using relatively computationally low-intensive processes, illustratively consisting of an anti-correlation process, and associated anti-correlation kernel, which operates upon a plurality of pixel neighborhoods within the image. This process enables an entire image to be quickly analyzed for any features that exhibit such saddle point-like structures by determining whether the anti-correlation kernel generates a weak or strong response in various positions within the image. The anti-correlation kernel is designed to generate a strong response regardless of the orientation of a saddle point-like structure. The anti-correlation process examines a plurality of pixel neighborhoods in the image, thereby locating any saddle point-like structures regardless of orientation, as it is angle-independent. The structures are then grouped and refined (for example in a grid) in an effort to locate and decode ID topologies within the image.

    Abstract translation: 本发明提供一种用于在图像中发现特征的系统和方法,该特征使用相对计算上的低密集过程表现出鞍点状结构,其示意性地由抗相关过程和相关联的反相关核组成,该相关反相内核基于多个像素 图像内的社区。 该过程使得能够通过确定反相关内核是否在图像内的各个位置中产生弱或强响应来呈现出诸如鞍点状结构的任何特征,从而快速分析整个图像。 反相关核被设计成产生强烈的响应,而不管鞍点状结构的取向如何。 反相关处理检查图像中的多个像素邻域,从而定位任何鞍状结构,而不管方位如何,因为它是与角度无关的。 然后将结构分组和精化(例如,在网格中),以努力定位和解码图像内的ID拓扑。

Patent Agency Ranking