No abstract

No abstract

With the recent development of multiprimary display devices, the three primary colors (red, green, and blue) of a conventional display need to be decomposed into control values for a multiprimary display (MPD) under the constraints of tristimulus matching. To achieve tristimulus matching between such different display systems, the MPD color signals need to be estimated based on a device-independent color space, such as CIEXYZ or CIELAB. Yet, since the focus of a MPD is to display motion picture data, the color space conversion and multiprimary control value decomposition must be simplified. Accordingly, this paper presents a color signal decomposition method for a MPD using a three-dimensional look-up-table (3D LUT) in linearized LAB space. Linearized LAB space satisfies the linearity and additivity required for the color space conversion, and can easily construct a 3D LUT that considers the lightness, chroma, and hue. In addition, to reproduce moving picture data in a MPD, the proposed decomposition method uses a 3D LUT structure to reduce the hardware complexity and processing time. First, a 3D LUT that contains the gamut boundary points of the MPD is created to decompose the multiprimary control values. The chroma and multiprimary color signals for the gamut boundary are then stored in the 3D LUT along with the quantized hue and lightness values. Next, a quadrangular pyramid composed of four gamut boundary points and one lightness point on an achromatic axis is generated according to the input linearized LAB values. Consequently, MPD color signals can be obtained for the input values by interpolating between the multiprimary color signals for the gamut boundary points and the lightness point on an achromatic axis. Furthermore, additional gamut boundary points within 10°of the hue are used to interpolate the input values in regions that involve an abrupt change in the multiprimary control values to achieve a smooth change of hue. As a result, the proposed method guarantees computational efficiency and color signal continuity. Plus, less memory space is required when compared with conventional color decomposition methods.

No abstract

Text-enhanced error diffusion is proposed to sharpen text regions in complex documents, including natural images with a multifunctional printer. To enhance the sharpness of the text regions, the input image is segmented into text and background regions using the maximum gradient difference. Edge-enhanced error diffusion is then applied to the text regions to sharpen the text, while Floyd and Steinberg's error diffusion is applied to the background regions to obtain smooth dot patterns. However, this combination of algorithms can generate two kinds of artifact around the text regions: boundary and dot-elimination artifacts. Boundary artifacts are a series of dots distributed around text blocks, and this propagation error generated below a text line by the edge-enhanced error diffusion is largely diffused forward into the background region. Thus, to gradually decrease these propagation errors, a grayscale dilation operator is processed along the boundary of a text block, thereby creating a gradual dilated transition region. Edge-enhanced error diffusion using different multiplicative parameters is then applied to these regions. Meanwhile, dot-elimination artifacts are dot-disappearing phenomena occurring around high-frequency regions due to the characteristic of the edge-enhanced error diffusion to sharpen edge regions more. Thus, an error scaling factor is inserted in front of the error filter in the architecture of the edge-enhanced error diffusion to scale down the propagation errors. Experiments demonstrate that text readability is improved by increasing the sharpness of the text regions, and a less grainy appearance is simultaneously achieved compared with conventional edge-enhanced error diffusion in the background regions.