Object recognition utilizing feature alignment

    公开(公告)号:US10204284B2

    公开(公告)日:2019-02-12

    申请号:US15666824

    申请日:2017-08-02

    Abstract: Systems, methods, and computer-readable storage media are provided for identifying (recognizing) an object from its shape in a sequence of images utilizing sequence alignment matrices (SAMs). For a given image, an object is segmented and from the segmented object, a set of key points is extracted. From the extracted key points, a set of local feature descriptors, strictly related to the key points and uniquely ordered in sequence, are extracted. The feature sequence obtained from the segmented object is aligned with a counterpart or reference image (e.g., a model or another image) using a Sequence Alignment Matrix (SAM). A custom scoring technique for the alignment provides a quality index for the identification of the object.

    SEGMENT-BASED PATTERN MATCHING ALGORITHM
    4.
    发明申请

    公开(公告)号:US20190244021A1

    公开(公告)日:2019-08-08

    申请号:US16271260

    申请日:2019-02-08

    CPC classification number: G06K9/00536 G06K9/00523

    Abstract: A system and method for matching a pattern may include generating a set of model descriptors representative of segment features of a first imaged object. A set of query descriptors representative of segment features of a second imaged object may be generated. A first model segment may be selected to align with the query segments to determine if any of the query segments correspond with the first model segment based on respective model descriptors and query descriptors. An hypothesis may be generated by computing a transformation to match the selected first model descriptor with a query descriptor. The hypothesis may be validated by a model fitting algorithm when comparing other transformed model descriptors with other query descriptors. Based on a consensus value from the model fitting algorithm, a determination as to whether a pattern match exists between the first and second images may be made.

    Describing objects using edge-pixel-feature descriptors

    公开(公告)号:US10095951B2

    公开(公告)日:2018-10-09

    申请号:US15351367

    申请日:2016-11-14

    Abstract: During a description technique, a local descriptor for an object may be generated by computing a 2-dimensional histogram of pairs of angles between pairs of line segments that are aligned with edge pixels associated with the object. The pairs of line segments belong to a subset of k neighboring or proximate line segments. Moreover, this 2D histogram may represent the relative displacement and the relative orientations of the pairs of line segments in the subset as weights in bins or cells defined by angular quantization values, and the 2D histogram may exclude lengths of the line segments. Subsequently, the generated 2D histogram may be compared to predefined sets of descriptors in a model library that are associated with a set of objects, and the object may be identified as one of the set of objects based on a group of match scores determined in the comparisons.

    Describing objects using edge-pixel-feature descriptors
    6.
    发明授权
    Describing objects using edge-pixel-feature descriptors 有权
    使用边缘像素特征描述符描述对象

    公开(公告)号:US09495607B2

    公开(公告)日:2016-11-15

    申请号:US14761967

    申请日:2013-01-21

    Abstract: During a description technique (100), a local descriptor for an object (300) is generated (122) by computing a 2-dimensional histogram (600) of pairs of angles (514, 516) between pairs of line segments (510, 512) that are aligned with edge pixels associated with the object (300). The pairs of line segments (510, 512) belong to a subset of k neighboring or proximate line segments (310). Moreover, this 2D histogram (600) may represent the relative displacement and the relative orientations of the pairs of line segments (510, 512) in the subset as weights in bins or cells defined by angular quantization values, and the 2D histogram (600 may exclude lengths of the line segments. Subsequently, the generated 2D histogram (600) may be compared (210) to predefined sets of descriptors in a model library that are associated with a set of objects, and the object may be identified (212) as one of the set of objects based on a group of match scores determined in the comparisons.

    Abstract translation: 在描述技术(100)期间,通过计算成对的线段(510,512)之间的成对角(514,516)的二维直方图(600)来生成对象(300)的局部描述符(122) ),其与与对象(300)相关联的边缘像素对齐。 线段对(510,512)属于k个相邻或邻近的线段(310)的子集。 此外,该2D直方图(600)可以表示子集中的线段对(510,512)的相对位移和相对取向,作为由角度量化值定义的小区或小区中的权重,并且2D直方图(600可以 随后,可以将所生成的2D直方图(600)(210)与与一组对象相关联的模型库中的预定义描述符集进行比较(210),并且可以将对象识别(212)为 基于在比较中确定的一组匹配得分的一组对象中的一个。

    IMAGING SYSTEM AND METHOD USING A MULTI-LAYER MODEL APPROACH TO PROVIDE ROBUST OBJECT DETECTION

    公开(公告)号:US20230038286A1

    公开(公告)日:2023-02-09

    申请号:US17394358

    申请日:2021-08-04

    Abstract: A system and method of detecting an image of a template object in a captured image may include comparing, by a processor, an image model of an imaged template object to multiple locations, rotations, and scales in the captured image. The image model may be defined by multiple model base point sets derived from contours of the imaged template object, where each model base point set inclusive of a plurality of model base points that are positioned at corresponding locations associated with distinctive features of the imaged template object. Each corresponding model base point of the model base point sets may (i) be associated with respective layers and (ii) have an associated gradient vector. A determination may be made as to whether and where the image of the object described by the image model is located in the captured image.

    Segment-based pattern matching algorithm

    公开(公告)号:US11113522B2

    公开(公告)日:2021-09-07

    申请号:US16271260

    申请日:2019-02-08

    Abstract: A system and method for matching a pattern may include generating a set of model descriptors representative of segment features of a first imaged object. A set of query descriptors representative of segment features of a second imaged object may be generated. A first model segment may be selected to align with the query segments to determine if any of the query segments correspond with the first model segment based on respective model descriptors and query descriptors. An hypothesis may be generated by computing a transformation to match the selected first model descriptor with a query descriptor. The hypothesis may be validated by a model fitting algorithm when comparing other transformed model descriptors with other query descriptors. Based on a consensus value from the model fitting algorithm, a determination as to whether a pattern match exists between the first and second images may be made.

    Object Recognition Utilizing Feature Alignment

    公开(公告)号:US20180157931A1

    公开(公告)日:2018-06-07

    申请号:US15666824

    申请日:2017-08-02

    CPC classification number: G06K9/4671 G06K9/50 G06K9/6211

    Abstract: Systems, methods, and computer-readable storage media are provided for identifying (recognizing) an object from its shape in a sequence of images utilizing sequence alignment matrices (SAMs). For a given image, an object is segmented and from the segmented object, a set of key points is extracted. From the extracted key points, a set of local feature descriptors, strictly related to the key points and uniquely ordered in sequence, are extracted. The feature sequence obtained from the segmented object is aligned with a counterpart or reference image (e.g., a model or another image) using a Sequence Alignment Matrix (SAM). A custom scoring technique for the alignment provides a quality index for the identification of the object.

Patent Agency Ranking