Showing papers presented at "Computational Color Imaging Workshop in 2009"

Spatio-temporal Tone Mapping Operator Based on a Retina Model

Abstract: From moonlight to bright sun shine, real world visual scenes contain a very wide range of luminance; they are said to be High Dynamic Range (HDR). Our visual system is well adapted to explore and analyze such a variable visual content. It is now possible to acquire such HDR contents with digital cameras; however it is not possible to render them all on standard displays, which have only Low Dynamic Range (LDR) capabilities. This rendering usually generates bad exposure or loss of information. It is necessary to develop locally adaptive Tone Mapping Operators (TMO) to compress a HDR content to a LDR one and keep as much information as possible. The human retina is known to perform such a task to overcome the limited range of values which can be coded by neurons. The purpose of this paper is to present a TMO inspired from the retina properties. The presented biological model allows reliable dynamic range compression with natural color constancy properties. Moreover, its non-separable spatio-temporal filter enhances HDR video content processing with an added temporal constancy.

A New Spatial Hue Angle Metric for Perceptual Image Difference

Abstract: Color image difference metrics have been proposed to find differences between an original image and a modified version of it. One of these metrics is the hue angle algorithm proposed by Hong and Luo in 2002. This metric does not take into account the spatial properties of the human visual system, and could therefore miscalculate the difference between an original image and a modified version of it. Because of this we propose a new color image difference metrics based on the hue angle algorithm that takes into account the spatial properties of the human visual system. The proposed metric, which we have named SHAME (Spatial Hue Angle MEtric), have been subjected to extensive testing. The results show improvement in performance compared to the original metric proposed by Hong and Luo.

Color Reproduction Using Riemann Normal Coordinates

Abstract: In this paper we use tools from Riemann geometry to construct color reproduction methods that take into account the varying color perception properties of observers. We summarize the different steps in the processing: the estimation of the discrimination thresholds, the estimation of the Riemann metric, the construction of the metric-preserving or color-difference-preserving mapping and the usage of Semantic Differentiation (SD) techniques in the evaluation. We tested the method by measuring the discrimination data for 45 observers. We illustrate the geometric properties of the color spaces obtained and show the result of the metric-preserving maps between the color space of color-normal observers and a color-weak observer. We then apply the obtained maps to produce color corrected reproductions of two natural images.

KANSEI Based Clothing Fabric Image Retrieval

Abstract: KANSEI is a Japanese term which means psychological feeling or image of a product. KANSEI engineering refers to the translation of consumers' psychological feeling about a product into perceptual design elements. Recently KANSEI based image indexing or image retrieval have been done by using interactive genetic algorithms (IGA). In this paper, we propose a new technique for clothing fabric image retrieval based on KANSEI (impressions). We first learn the mapping function between the fabric image features and the KANSEI and then the images in the database are projected into the KANSEI space (psychological space). The retrieval is done in the psychological space by comparing the query impression with the projection of the images in database.

Material Classification for Printed Circuit Boards by Spectral Imaging System

Abstract: This paper presents an approach to a reliable material classification for printed circuit boards (PCBs) by constructing a spectral imaging system. The system works in the whole spectral range [400-700nm] and the high spectral resolution. An algorithm is presented for effectively classifying the surface material on each pixel point into several elements such as substrate, metal, resist, footprint, and paint, based on the surface-spectral reflectance estimated from the spectral imaging data. The proposed approach is an incorporation of spectral reflectance estimation, spectral feature extraction, and image segmentation processes for material classification of raw PCBs. The performance of the proposed method is compared with other methods using the RGB-reflectance based algorithm, the k-means algorithm and the normalized cut algorithm. The experimental results show the superiority of our method in accuracy and computational cost.

Non-linear Filter Response Distributions of Natural Colour Images

Abstract: We observe a non-Gaussian heavy tailed distribution for the non-linear filter 1$$ \label{Filter} \gamma(U)(\mathbf r) = U(\bold r) - \sum_{ \mathbf s \in N(\mathbf r)} w{(Y)_{\mathbf r \mathbf s}} U(\mathbf s), $$applied to the chromacity channel 'U' (and equivalently to 'V') on individual natural colour images in the colour space YUV. We fit a Generalised Gaussian Distribution (GGD) to the histogram of the filter response, and observe the shape parameter (?) to lie within the range 0 1.

An Adaptive Tone Mapping Algorithm for High Dynamic Range Images

Abstract: Real world scenes contain a large range of light intensities which range from dim starlight to bright sunlight. A common task of tone mapping algorithms is to reproduce high dynamic range(HDR) images on low dynamic range(LDR) display devices such as printers and monitors. In this paper, a new tone mapping algorithm is proposed for the display of HDR images. Inspired by the adaptive process of the human visual system, the proposed algorithm utilized the center-surround Retinex processing. The novelty of our method is that the local details are enhanced according to a non-linear adaptive spatial filter (Gaussian filter), whose shape is adapted to high-contrast edges of the image. The proposed method uses an adaptive surround instead of the traditional pre-defined circular. Therefore, the algorithm can preserve visibility and contrast impression of high dynamic range scenes in the common display devices. The proposed method is tested on a variety of HDR images, and we also compare it to previous work. The results show good performance of our method in terms of visual quality.

Colour Representation in Lateral Geniculate Nucleus and Natural Colour Distributions

Abstract: We investigated the representation of a wide range of colours in the lateral geniculate nucleus (LGN) of macaque monkeys We took an approach to reconstruct a colour space from responses of a population of neurons We found that, in the derived colour space (`LGN colour space'), red and blue regions were compressed whereas purple region was expanded, compared with those in a linear cone-opponent colour space We found that the expanding/compressing pattern in the LGN colour space was related to the colour histogram derived from a natural image database Quantitative analysis showed that the response functions of the population of the neurons were nearly optimal according to the principle of 'minimizing errors in estimation of stimulus colour in the presence of response noise' Our findings support the idea that the colour representation at the early neural processing stage is adapted for efficient coding of colour information in the natural environment

Perceptual Color Correction: A Variational Perspective

Abstract: Variational techniques provide a powerful tool for understanding image features and creating new efficient algorithms. In the past twenty years, this machinery has been also applied to color images. Recently, a general variational framework that incorporates the basic phenomenological characteristics of the human visual system has been built. Here we recall the structure of this framework and give noticeable examples. We then propose a new analytic expression for a parameter that regulates contrast enhancement. This formula is defined in terms of intrinsic image features, so that the parameter no longer needs to be empirically set by a user, but it is automatically determined by the image itself.

Texture Sensitive Denoising for Single Sensor Color Imaging Devices

Abstract: This paper presents a spatial noise reduction technique designed to work on CFA (Color Filter Array) data acquired by CCD/CMOS image sensors. The overall processing preserves image details by using heuristics related to HVS (Human Visual System) and texture detection. The estimated amount of texture and HVS sensitivity are combined to regulate the filter strength. Experimental results confirm the effectiveness of the proposed technique.

Illuminant Change Estimation via Minimization of Color Histogram Divergence

Abstract: We present a new method for computing the change of light possibly occurring between two pictures of the same scene We approximate the illuminant variation with the von Kries diagonal transform and estimate it by minimizing a functional that measures the divergence between the image color histograms Our approach shows good performances in terms of accuracy of the illuminant change estimation and of robustness to pixel saturation and Gaussian noise Moreover we illustrate how the method can be applied to solve the problem of illuminant invariant image recognition

Illumination Chromaticity Estimation Based on Dichromatic Reflection Model and Imperfect Segmentation

Abstract: The illumination chromaticity estimation based on the dichromatic reflection model has not been made practicable, since the method needs image segmentation beforehand. However, its two-dimensional model is sufficiently robust, when it is combined with the least square method. The proposed algorithm executes the color space division instead of the segmentation. The original image is divided into small color regions, each of which corresponds to one of color sub-spaces. Though this division is imperfect image segmentation, the illumination chromaticity estimation based on the chromaticity distribution in the color regions is possible. Experimental result shows that this method is also applicable to images of apparently matt surfaces.

A Colorimetric Study of Spatial Uniformity in Projection Displays

Abstract: In this paper we investigate and study the color spatial uniformity of projectors. A common assumption is to consider that only the luminance is varying along the spatial dimension. We show that the chromaticity plays a significant role in the spatial color shift, and should not be disregarded, depending on the application. We base our conclusions on the measurements obtained from three projectors. Two methods are used to analyze the data, a conventional approach, and a new one which considers 3D gamut differences. The results show that the color gamut difference between two spatial coordinates within the same display can be larger than the difference observed between two projectors.

Color Constancy Algorithm Selection Using CART

Abstract: In this work, we investigate how illuminant estimation techniques can be improved taking into account intrinsic, low level properties of the images. We show how these properties can be used to drive, given a set of illuminant estimation algorithms, the selection of the best algorithm for a given image. The selection is made by a decision forest composed by several trees that vote for one of the illuminant estimation algorithm. The most voted algorithm is then applied to the input image. Experimental results on the widely used Ciurea and Funt dataset demonstrate the accuracy of our approach in comparison to other algorithms in the state of the art.

Structure Tensor of Colour Quaternion Image Representations for Invariant Feature Extraction

Abstract: Colour image representation using real quaternions has shown to be very useful for linear and morphological colour filtering. This paper deals with the extension of first derivatives-based structure tensor for various quaternionic colour image representations. Classical corner and edge features are obtained from eigenvalues of the quaternionic colour structure tensors. We study the properties of invariance of the quaternion colour spatial derivatives and their robustness for feature extraction on practical examples.

Polynomial Regression Spectra Reconstruction of Arctic Charr's RGB

Abstract: Arctic Charr (Salvelinus alpinus L.) exhibit red ornamentation at abdomen area during the mating season. The redness is caused by carotenoid components and it assumed to be related to the vitality, nutritional status, foraging ability and generally health of the fish. To assess the carotenoid amount, the spectral data is preferred but it is not always possible to measure it. Therefore, an RGB-to-spectra transform is needed. We test here polynomial regression model with different training sets to find good model especially for Arctic charr.

Color Information Processing in Higher Brain Areas

Abstract: Significant color signal transformation occurs in the primary visual cortex and neurons tuned to various direction in the color space are generated. The resulting multi-axes color representation appears to be the basic principle of color representation throughout the visual cortex. Color signal is conveyed through the ventral stream of cortical visual pathway and finally reaches to the inferior temporal (IT) cortex. Lesion studies have shown that IT cortex plays critical role in color vision. Color discrimination is accomplished by using the activities of a large number of color selective IT neurons with various properties. Both discrimination and categorization are important aspects of our color vision, and we can switch between these two modes depending on the task demand. IT cortex receives top-down signal coding the task and this signal adaptively modulates the color selective responses in IT cortex such that neural signals useful for the ongoing task is efficiently selected.

An Improved Image Re-indexing Technique by Self Organizing Motor Maps

Abstract: The paper presents a novel Motor Map neural network for re-indexing color mapped images. The overall learning process is able to smooth the local spatial redundancy of the indexes of the input image. Differently than before, the proposed optimization process is specifically devoted to re-organize the matrix of differences of the indexes computed according to some predefined patterns. Experimental results show that the proposed approach achieves good performances both in terms of compression ratio and zero order entropy of local differences. Also its computational complexity is competitive with previous works in the field.

Classification of Paper Images to Predict Substrate Parameters Prior to Print

Abstract: An accurate characterization of the substrate is a prerequisite of color management in print. The use of standard ICC profiles in prepress leaves it to the printer to match the fixed substrate characteristics contained in these profiles. This triggers the interest in methods to predict, if a given ink, press and paper combination complies with a given characterization. We present an approach to compare physical and optical characteristics of papers in order to achieve such a prediction of compliance by classification methods. For economical and ecological reasons it is preferable to test paper without printing it. We therefore propose non-destructive methods.

Color Stereo Matching Cost Applied to CFA Images

Abstract: Most color stereovision setups include single-sensor cameras which provide Color Filter Array (CFA) images. In those, a single color component is sampled at each pixel rather than the three required ones (R,G,B). We show that standard demosaicing techniques, used to interpolate missing components, are not well adapted when the resulting color pixels are matched for estimating image disparities. In order to avoid this problem while exploiting color information, we propose a new matching cost designed for dense stereovision based on pairs of CFA images.

Synthesis of Facial Images with Foundation Make-Up

Abstract: A method is described for synthesizing color images of whole human face with foundation make-up by using bare face image and the surface-spectral reflectance of the bare cheek. The synthesis of made-up facial images is based on the estimation of skin color with foundation make-up and the control of skin texture. First, the made-up skin color is estimated from the spectral reflectance of the bare cheek and the optical properties of the foundation. The spectral reflectance of made-up skin is calculated by the Kubelka-Munk theory. Second, smooth texture control is done by the intensity change of layers in the multi-resolution analysis with the Daubechies wavelet. Luster is enhanced and acnes are attenuated by the texture control. Experimental results show the accurate estimation of made-up skin color and the effective texture control. It is shown that the made-up face images are rendered with sufficient accuracy.

A Computationally Efficient Technique for Image Colorization

Abstract: In this paper, the fast technique for image colorization is considered. The proposed method transfers colors from the color image (source) to the gray level image (target). For the source image, we use the segmented uniformly colored regions (dielectric surfaces) under single color illumination. This method maps the gray level image into the color space by means of parametrical mapping learnt using PCA and principal components regression. The experiments show the method's feasibility for colorizing the objects, and textures, as well.

Supervised Local Subspace Learning for Region Segmentation and Categorization in High-Resolution Satellite Images

Abstract: We proposed a new feature extraction method based on supervised locality preserving projections (SLPP) for region segmentation and categorization in high-resolution satellite images. Compared with other subspace methods such as PCA and ICA, SLPP can preserve local geometric structure of data and enhance within-class local information. The generalization of the proposed SLPP based method is discussed in this paper.

JBIG for Printer Pipelines: A Compression Test

Abstract: The proposed paper describes a compression test analysis of JBIG standard algorithm. The aim of such work is to proof the effectiveness of this standard for images acquired through scanners and processed into a printer pipeline. The main issue of printer pipelines is the necessity to use a memory buffer to store scanned images for multiple prints. This work demonstrates that for very large scales the buffer can be fixed using medium compression case, using multiple scans in case of uncommon random patterns.