A Study of Color Image Segmentation Using Moment-Preserving Technigue

• Main Authors: Lin, Yung Jehn, 林永鎮
• Other Authors: Chien-Pen Chuang
• Format: Others
• Language: zh-TW
• Published: 1996
• Online Access: http://ndltd.ncl.edu.tw/handle/30347725602698445621

碩士 === 國立臺灣師範大學 === 工業教育學系 === 84 === The purpose of this study is to investigate the effect of colorsegmentation using moment-preserving technique.it is based on the human eyes'' chatacteristics of color clustering which was discoveredby MacAdam in 1943. According to MacAdam''s ellipses, human eyes cannot recognize neighbor colors within limited difference even though those colors located at different coordinates of CIE colormap. It means that our eyes'' color sensitivity for different coloris different and can be figured as clusters in MacAdam''s ellipses.Among three primary colors, blue is the most sensitive color for human eyes and then goes red and green. But the LUT method which is considered as the most precise color segmentation method cannot segment colorsas our natural eyes do. The output image of LUT method contain the pixelswith only one selected RGB value that is not available for human visionin some cases, especially for the segmentation of the colors with thesame hue but different saturation or different brightness. In order tosimulate human eyes''color segmentation function, a three order moment-preserving technique is used to preserve sevral neighbor RGB value within two threshold values . If the cursor is pointed to the color pixel which color is closed to green, then a 9×9 mask is used to calculatethe momentum of RGB values.And then output the image which RGB valuesare within the calculated threshold values. If the selected pixel which color is closed to red or blue, a 5×5 or 3×3 mask is used to do momentpreserving processing separately.