公开(公告)号：US08326030B2

公开(公告)日：2012-12-04

申请号：US13063334

申请日：2009-09-17

申请人： Rikio Onai ,  Takahiro Hayashi ,  Tatsuya Kiyono ,  Masaya Mori ,  Masahiro Sanjo

发明人： Rikio Onai ,  Takahiro Hayashi ,  Tatsuya Kiyono ,  Masaya Mori ,  Masahiro Sanjo

IPC分类号： G06K9/00

摘要： There are provided a foreground region extraction program, a foreground region extraction apparatus, and a foreground region extraction method capable of clipping an object region (foreground region) at high speed.An arrangement is as follows: Foreground pixels designated by a user and background pixels designated thereby are each set as a reference pixel to identify, from among a plurality of divided color spaces created by a three-dimensional color space being divided, the divided color space to which each of the reference pixels belongs, as a reference divided color space. Color distance calculation processing to calculate a color distance in the color space between each reference pixel and an adjacent pixel adjacent thereto is performed, and belonging judgment processing to judge whether each of the adjacent pixels belongs to each of the reference divided color spaces is performed, and then, cost calculation processing to calculate a cost of each adjacent pixel is performed, based on the color distance calculated for each of the adjacent pixels, and weights based on whether to belong to the reference divided color space judged for each of the adjacent pixels. Further determination processing to determine the adjacent pixel having the lowest calculated cost as a foreground pixel or background pixel is performed, and a foreground region is extracted from the image by using the determined adjacent pixel as a reference pixel to repeatedly perform the color distance calculation processing, the belonging judgment processing, the cost calculation processing, and the determination processing.

公开(公告)号：US08660309B2

公开(公告)日：2014-02-25

申请号：US13502972

申请日：2010-10-18

申请人： Rikio Onai ,  Tomohiro Sato ,  Masaya Mori ,  Masahiro Sanjo

发明人： Rikio Onai ,  Tomohiro Sato ,  Masaya Mori ,  Masahiro Sanjo

IPC分类号： G06K9/00

CPC分类号： H04N1/3878 ,  G06T3/00

摘要： There are provided an image processing apparatus, an image processing method and an image processing program for transforming a target image having no contour of straight line portions. An ellipse E1 by which the target image is approximated is calculated (S20), a center candidate O.sub.A of the target image is decided (S21), a first polygon P1 associated with the ellipse and the center candidate is calculated (S22), a projective transformation matrix for projectively transforming the first polygon into a second polygon P1′ associated with an ellipse whose center is a point where the center candidate is projectively transformed is calculated (S23), and the target image is projectively transformed based on the projective transformation matrix to obtain a transformation image (S24).

摘要翻译： 提供了一种图像处理装置，图像处理方法和用于变换没有直线部分轮廓的目标图像的图像处理程序。 计算近似目标图像的椭圆E1（S20），确定目标图像的中心候选OA（S21），计算与椭圆和中心候选相关联的第一多边形P1（S22 ），投影变换矩阵，用于将第一多边形投影变换成与其中心是投影变换中心候选点的中心的椭圆相关联的第二多边形P1'（S23），并且基于目标图像进行投影变换 投影变换矩阵以获得变换图像（S24）。

公开(公告)号：US20120201423A1

公开(公告)日：2012-08-09

申请号：US13502972

申请日：2010-10-18

申请人： Rikio Onai ,  Tomohiro Sato ,  Masaya Mori ,  Masahiro Sanjo

发明人： Rikio Onai ,  Tomohiro Sato ,  Masaya Mori ,  Masahiro Sanjo

IPC分类号： G06K9/48 ,  G06K9/00

CPC分类号： H04N1/3878 ,  G06T3/00

摘要： There are provided an image processing apparatus, an image processing method and an image processing program for transforming a target image having no contour of straight line portions.An ellipse E1 by which the target image is approximated is calculated (S20), a center candidate OA of the target image is decided (S21), a first polygon P1 associated with the ellipse and the center candidate is calculated (S22), a projective transformation matrix for projectively transforming the first polygon into a second polygon P1′ associated with an ellipse whose center is a point where the center candidate is projectively transformed is calculated (S23), and the target image is projectively transformed based on the projective transformation matrix to obtain a transformation image (S24).

摘要翻译： 提供了一种图像处理装置，图像处理方法和用于变换没有直线部分轮廓的目标图像的图像处理程序。 计算近似目标图像的椭圆E1（S20），确定目标图像的中心候选OA（S21），计算与椭圆和中心候选相关联的第一多边形P1（S22），投影 转换矩阵，用于将第一多边形投影变换为与其中心是中心候选投影变换的点相关联的第二多边形P1'（S23），并且基于投影变换矩阵对目标图像进行投影变换 获得变换图像（S24）。

公开(公告)号：US20110164814A1

公开(公告)日：2011-07-07

申请号：US13063334

申请日：2009-09-17

申请人： Rikio Onai ,  Takahiro Hayashi ,  Tatsuya Kiyono ,  Masaya Mori ,  Masahiro Sanjo

发明人： Rikio Onai ,  Takahiro Hayashi ,  Tatsuya Kiyono ,  Masaya Mori ,  Masahiro Sanjo

IPC分类号： G06K9/34

CPC分类号： G06T7/11 ,  G06T7/187 ,  G06T7/194 ,  G06T2207/10024 ,  G06T2207/20101

摘要： There are provided a foreground region extraction program, a foreground region extraction apparatus, and a foreground region extraction method capable of clipping an object region (foreground region) at high speed.An arrangement is as follows: Foreground pixels designated by a user and background pixels designated thereby are each set as a reference pixel to identify, from among a plurality of divided color spaces created by a three-dimensional color space being divided, the divided color space to which each of the reference pixels belongs, as a reference divided color space. Color distance calculation processing to calculate a color distance in the color space between each reference pixel and an adjacent pixel adjacent thereto is performed, and belonging judgment processing to judge whether each of the adjacent pixels belongs to each of the reference divided color spaces is performed, and then, cost calculation processing to calculate a cost of each adjacent pixel is performed, based on the color distance calculated for each of the adjacent pixels, and weights based on whether to belong to the reference divided color space judged for each of the adjacent pixels. Further determination processing to determine the adjacent pixel having the lowest calculated cost as a foreground pixel or background pixel is performed, and a foreground region is extracted from the image by using the determined adjacent pixel as a reference pixel to repeatedly perform the color distance calculation processing, the belonging judgment processing, the cost calculation processing, and the determination processing.

摘要翻译： 提供了能够高速地削减对象区域（前景区域）的前景区域提取程序，前景区域提取装置和前景区域提取方法。 布置如下：由用户指定的前景像素和由此指定的背景像素分别设置为参考像素，以识别由被划分的三维颜色空间创建的多个分割颜色空间中的分割颜色空间 每个参考像素所属的，作为参考分色色空间。 执行用于计算每个参考像素和与其相邻的相邻像素之间的颜色空间中的颜色距离的颜色距离计算处理，以及属性判断处理，以判断每个相邻像素是否属于每个参考分割色彩空间， 然后，基于为每个相邻像素计算出的颜色距离和基于是否属于为每个相邻像素判断的参考分割颜色空间的权重来执行用于计算每个相邻像素的成本的成本计算处理 。 执行将具有最低计算成本的相邻像素确定为前景像素或背景像素的进一步确定处理，并且通过使用所确定的相邻像素作为参考像素从图像中提取前景区域以重复执行色彩距离计算处理 归属判断处理，成本计算处理和确定处理。