画像解析による水玉模様の同時対比効果の新しい評価法

概要

論文の詳細を見る
Image analysis is applied to the characterization of color contrast of polka-dotted patterns. The color system developed in this research permit image features to be identified not only by lightness but also by other color attributes. Colors in an image are expressed in the NTSC_RGB color system. Therefore, three primary colors, red, green and blue (RGB), can be obtained for each pixel directly from the image. However, the RGB color system does not directly correspond to a person's visual perceptions. On the other hand, a color coordinate system based on hue (H^*), chroma (C^*) and lightness (L^*) is more consisted with visual perceptions. Consequently, RGB images are converted to L^*C^*H^* images. Each pixel of the images is assigned a gray-level value from 0 for black to 255 for white. The co-occurrence contrast (CON) is calculated for L^*, C^* and H^* images. CON for achromatic images with a lightness contrast effect increases with the increase of the difference of lightness between the polka-dotted pattern (the test field) and the background (the inducing field), regardless of whether the lightness of the figure becomes higher or lower than that of the background. The simultaneous contrast effect combined with lightness, saturation and hue contrast is also discussed in terms of CON.
2006-02-15