Color Space Transformation from RGB to CIELAB Using Neural Networks
PCM '09 Proceedings of the 10th Pacific Rim Conference on Multimedia: Advances in Multimedia Information Processing
| Hi-index | 0.43 |
Low priced consumer video devices have become increasingly common. Furthermore, with the promotion of digital and multimedia devices, the pictures obtained from such devices are not only subjected to printing, display and editing, but also are transmitted through network systems. However, the transmission of color information involves the problem of color developments which are different in type, calling for the study of the transmission of color information in the standard color space, independent of the device. By means of the conventional system, which jointly uses a low-capacity 3-dimensional LUT (look up table) and linear interpolation, it is not possible to improve the transformation precision unless the capacity of the 3-dimensional LUT is increased. Therefore the present study investigates upgrading the transformation precision such as by expanding the transformation region, correcting the color displacement, using a 3-dimensional LUT with the output values arranged at equal intervals of space, and improving the dark section. This has led to the circuit size slightly increasing due to the added 3-dimensional LUT and the need for dividers; excellent color transformation is assured. The color difference between the original image and the reproduced image after repeating the transformation ten times can be reduced to 30% of the conventional system, providing excellent results in terms of the subjective estimation of gradation continuity, color displacement, etc