A method for performing color reduction in index images
Date Issued
2004
Date
2004
Author(s)
Lo, Shih Chieh
DOI
en-US
Abstract
In conventional color reduction for indexed color image, it is necessary to convert into a full color image. Then, performing color quantization with the full color image to generate a new palette with a predetermined number of colors.
However, a large memory space is needed to temporally store the full color image, and the color quantization process with the full color image is complicated and time-consuming. The computing time depends on the dimension of image. The bigger size the image owns, the more time the analysis needs. Additionally, when more than one source image frames in one image file use a same palette, different new palettes for each frame are generated after performing the conventional color reduction. This side effect will cause all frame of the image cannot share one palette any more. The storage usage of the image is inefficient.
In this paper, we will propose a method to reduce the color used in an indexed color image. The method will make the color reduce process more efficient and with lower cost. Compare with conventional method, the memory usage of our method is limited. The performance is also better than before. Besides, our method also solves the multiple frames in one image issue. Regarding the benefits, the real time or some heavy-using environment, like transcoding server, can utilize this method to fulfill their requirement. With the compression enabled file formats, such as GIF and PNG file, it will cause the higher compression rate and also make the lower transmitting cost.
Subjects
調色盤
影像轉換
減色
索引圖
Color reduction
Palette
Image transcoding
Index color
Type
thesis