@article{bbc26c818f24496894f56bdda9f0ce49,
title = "Effects of Luminance Quantization Error on Color Image Processing",
abstract = "A common approach to color image processing is to apply monochrome techniques to the quantized intensity component. However, the quantization error in the intermediate intensity image propagates to the final processed image. This can lead to significant distortion, depending on the input RGB values and on the particular image processing function being applied. A theoretical analysis of the worst-case quantization error is presented. Experimental results with histogram equalization demonstrate how the histogram resolution affects the performance.",
author = "Rodr{\'i}guez, {Jeffrey J.} and Yang, {Christopher C.}",
note = "Funding Information: I. INTRODUCTION In color image processing, a traditional approach is to apply a coordinate transformation to the RGB image to obtain the luminous intensity component. Monochrome image processing techniques can then be applied to the luminance, or intensity, image. Finally, an inverse coordinate transformation is applied, resulting in the processed image. In fact, many applications of color image processing only require modification of the intensity component. For example, filtering, enhancement, restoration, and edge detection are often applied only to the intensity component of color images. In such cases, the other transformed coordinate components are not processed and do not even need to be computed. To reduce memory and disk space requirements, the intensity value for each pixel is often quantized to one of 256 discrete levels, which can be represented by an 8-bit binary number. The quantized intensity values are then processed to yield modified intensities, which are also quantized to 8-bit values. The RGB pixel values of the processed image are then computed by inverse coordinate transformation, or an equivalent procedure. The final RGB values are quantized to 8-bit values. Figs. 1 and 2 illustrate the stages of this process, with and without intermediate intensity quantization. Unfortunately, the quantization of the intermediate intensity values can lead to distortion in the output RGB values. Several researchers have studied the effects of quantization error in monochrome and color images. They have also derived methods to determine the minimum number of bits for quantization. Sezan, et al. [ 101 analytically and experimentally investigated the effects of uniform quantization in perceived lightness on the visibility of quantization noise in the context of digital radiography. A logarithmic model, several power function models, cone models, and a linear model were used to calculate the incremental luminance contrast as a function of density for different bit depths. Gentile, et al. [2] investigated algorithms based on image-independent and image-dependent quantizers implemented in RGB and CIELUV color spaces for digital printing and display of color images at near original quality. Two halftoning techniques, error diffusion and ordered dither, were used to reduce the number of output colors to achieve near original image quality. Ikeda, et al. [4] studied the perceptual colorimetric errors caused by quantizing color information expressed in XYZ and Manuscript received July 8, 1992; revised October 13, 1993. This work was supported by National Science Foundation Grant IRI-9112350. The associate editor coordinating the review of this paper and approving it for publication was Dr. M. Ibrahim Sezan. The authors are with the Department of Electrical and Computer Engineering, University of Arizona, Tucson, AZ 85721 USA. IEEE Log Number 9404452.",
year = "1994",
month = nov,
doi = "10.1109/83.336254",
language = "English (US)",
volume = "3",
pages = "850--854",
journal = "IEEE Transactions on Image Processing",
issn = "1057-7149",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
number = "6",
}