公开(公告)号：US08385642B2

公开(公告)日：2013-02-26

申请号：US12767516

申请日：2010-04-26

Applicant: Che-Lun Chuang ,  Chung-Ta Wu

Inventor： Che-Lun Chuang ,  Chung-Ta Wu

IPC: G06K9/36

CPC classification number: H04N1/58

Abstract: A method for removing color fringe is presented. Separated luminance and chrominance (YCbCr) signals of a digital image are analyzed through specific color detection, luminance detection, and gradient color detection, so as to determine whether color fringe occurs to each pixel in the digital image, thereby correcting pixels with color fringe.

Abstract translation: 提出了一种去除颜色条纹的方法。 通过特定的颜色检测，亮度检测和梯度颜色检测来分析数字图像的分离的亮度和色度（YCbCr）信号，以便确定是否对数字图像中的每个像素发生颜色条纹，由此校正具有彩色条纹的像素。

公开(公告)号：US08379977B2

公开(公告)日：2013-02-19

申请号：US12767553

申请日：2010-04-26

Applicant: Che-Lun Chuang ,  Chung-Ta Wu

Inventor： Che-Lun Chuang ,  Chung-Ta Wu

IPC: G06K9/36

CPC classification number: H04N1/58

Abstract: A method for removing color fringe is presented. First, detection and correction of color fringe are performed on an original-size image and a small-size image respectively, so as to generate respective corrected images and corresponding color fringe maps. Then, the corrected small-size image and its corresponding color fringe map are enlarged to the same resolution as the original-size image. Finally, the two corrected images are blended according to the respective corresponding color fringe maps.

Abstract translation: 提出了一种去除颜色条纹的方法。 首先，分别对原稿尺寸图像和小尺寸图像执行彩色条纹的检测和校正，以便产生相应的校正图像和对应的颜色条纹图。 然后，校正后的小尺寸图像及其相应的颜色条纹图被放大到与原始尺寸图像相同的分辨率。 最后，根据各自对应的颜色条纹图来混合两个校正图像。

公开(公告)号：US08270715B2

公开(公告)日：2012-09-18

申请号：US12339723

申请日：2008-12-19

Applicant: Chung-Fang Chien ,  Che-Lun Chuang

Inventor： Chung-Fang Chien ,  Che-Lun Chuang

IPC: G06K9/00

CPC classification number: G06K9/0061

Abstract: A method for correcting red-eye is described. Through facial features, at least one facial region is obtained in an image, a nose position in each facial region is obtained by using a nose feature, and at least one eye position is obtained based on a relative position relation between the nose and the eyes. After a color gamut of the image is converted, a red region is obtained from the eye position, and a plurality of edges is formed by using a luminance of the color gamut on the image with the converted color gamut according to the eye feature, so as to exclude the red region out of the plurality of edges, thereby improving accuracy of the red region on the eye position. Then, the red region is covered by an iris color, so as to correct the red-eye.

Abstract translation: 描述了一种用于校正红眼的方法。 通过面部特征，在图像中获得至少一个面部区域，通过使用鼻部特征获得每个面部区域中的鼻部位置，并且基于鼻子和眼睛之间的相对位置关系获得至少一个眼睛位置 。 在图像的色域被转换后，从眼睛位置获得红色区域，并且通过使用根据眼睛特征的具有转换色域的图像上的色域的亮度来形成多个边缘，因此 以排除多个边缘之外的红色区域，从而提高眼睛位置上的红色区域的精度。 然后，红色区域被虹膜颜色覆盖，以便校正红眼。

公开(公告)号：US20090161951A1

公开(公告)日：2009-06-25

申请号：US12339723

申请日：2008-12-19

Applicant: Chung-Fang Chien ,  Che-Lun Chuang

Inventor： Chung-Fang Chien ,  Che-Lun Chuang

IPC: G06K9/00

CPC classification number: G06K9/0061

Abstract: A method for correcting red-eye is described. Through facial features, at least one facial region is obtained in an image, a nose position in each facial region is obtained by using a nose feature, and at least one eye position is obtained based on a relative position relation between the nose and the eyes. After a color gamut of the image is converted, a red region is obtained from the eye position, and a plurality of edges is formed by using a luminance of the color gamut on the image with the converted color gamut according to the eye feature, so as to exclude the red region out of the plurality of edges, thereby improving accuracy of the red region on the eye position. Then, the red region is covered by an iris color, so as to correct the red-eye.

Abstract translation: 描述了一种用于校正红眼的方法。 通过面部特征，在图像中获得至少一个面部区域，通过使用鼻部特征获得每个面部区域中的鼻部位置，并且基于鼻子和眼睛之间的相对位置关系获得至少一个眼睛位置 。 在图像的色域被转换后，从眼睛位置获得红色区域，并且通过使用根据眼睛特征的具有转换色域的图像上的色域的亮度来形成多个边缘，因此 以排除多个边缘之外的红色区域，从而提高眼睛位置上的红色区域的精度。 然后，红色区域被虹膜颜色覆盖，以便校正红眼。