公开(公告)号：US20100202693A1

公开(公告)日：2010-08-12

申请号：US12658322

申请日：2010-02-08

申请人： Hee-Jun Song ,  Jae-Han Yoon ,  Hyun-Sik Shim ,  Young-Hee Park

发明人： Hee-Jun Song ,  Jae-Han Yoon ,  Hyun-Sik Shim ,  Young-Hee Park

IPC分类号： G06K9/46

CPC分类号： G06K9/00375 ,  G06K9/6256 ,  G06K2009/4666

摘要： A portable terminal is configured to recognize an image such as a hand shape. The portable terminal includes a motion detection unit that includes a trainer capable of generating uniform training image data to collectively generate a plurality of images of a desired region obtained from an original image in an identical posture and an identical size, and performing a training process to prevent a feature point from being generated in a portion where detection is unnecessary in the generated training image data, and a detector capable of detecting a rotated object from input data, after the training process.

摘要翻译： 便携式终端被配置为识别诸如手形的图像。 便携式终端包括运动检测单元，其包括能够产生均匀训练图像数据的训练器，以相同的姿势和相同的大小共同地生成从原始图像获得的期望区域的多个图像，并且执行训练处理 防止在生成的训练图像数据中检测不必要的部分中产生特征点，以及在训练处理之后能够从输入数据检测旋转物体的检测器。

公开(公告)号：US09111137B2

公开(公告)日：2015-08-18

申请号：US12658322

申请日：2010-02-08

申请人： Hee-Jun Song ,  Jae-Han Yoon ,  Hyun-Sik Shim ,  Young-Hee Park

发明人： Hee-Jun Song ,  Jae-Han Yoon ,  Hyun-Sik Shim ,  Young-Hee Park

IPC分类号： G06K9/62 ,  G06K9/00 ,  G06K9/46

CPC分类号： G06K9/00375 ,  G06K9/6256 ,  G06K2009/4666

摘要： A portable terminal is configured to recognize an image such as a hand shape. The portable terminal includes a motion detection unit that includes a trainer capable of generating uniform training image data to collectively generate a plurality of images of a desired region obtained from an original image in an identical posture and an identical size, and performing a training process to prevent a feature point from being generated in a portion where detection is unnecessary in the generated training image data, and a detector capable of detecting a rotated object from input data, after the training process.

摘要翻译： 便携式终端被配置为识别诸如手形的图像。 便携式终端包括运动检测单元，其包括能够产生均匀训练图像数据的训练器，以相同的姿势和相同的大小共同地生成从原始图像获得的期望区域的多个图像，并且执行训练处理 防止在生成的训练图像数据中检测不必要的部分中产生特征点，以及在训练处理之后能够从输入数据检测旋转物体的检测器。