For a practical measurement of sensitivity metamerism index for digital still camera, we have proposed Camera Rendering Index (CRI) using color patches defined in CIE 13.3. We evaluated it from several aspects in this paper. First, based on 13 types of sensitivity sets including real camera sensitivity, correlations between sensitivity metamerism indexes and color error for the virtual representation of typical objects are investigated. CRI gives the best correlation among some sensitivity metamerism indexes in terms of linearity with potential color errors caused by metamerism. Second, we discuss CRI with some modified parameters and procedures such as choice of color patches, spectral deviations of color patches, evaluation of Special CRI, color difference formula, optimization of color matrix, and final index formula. We found that the choice of the CIE 13.3 color patches is reasonable, and their spectral characteristics are critical while the other parameters and procedures may be insignificant. Base on these evaluations, we are proposing this evaluation index for an ISO standard.

This paper shows the results of the application of S-CIELAB colour metric to digital images of the fragments coming from the S. Mathew’s fresco, in the S. Francis Upper Church in Assisi, broken during the earthquake of September 1997. S-CIELAB, a spatial extension to the CIELAB colour representation, is a colour space whose metrics closely reproduce, through Euclidean norm, the colour distances perceived by a human observer. S-CIELAB accounts for the effects due to the spatial distribution of colours into the image. It is based on the application of a suitable spatial filtering to the colour image data, aimed to simulate the spatial blurring measured on the human visual system. In the system for aided virtual recomposition of fragmented frescos we are developing, the interaction between the operators and the huge collection of fragments is grounded on a ‘query-by-example’ paradigm. On the base of a set of images of fragments provided as examples, the system browses through the whole collection and retrieves all fragments similar in terms of their colour and texture contents. The application of S-CIELAB colour metric allows the system to ascribe to each fragment its pattern-colour characteristics according to the visual perception of the human operators. The close approximation of human perception allows a meaningful co-operation, increasing the system’s efficiency in the retrieval of fragments and reducing the workload of the restorer.

In this paper,we propose an automatic image processing system to beautify human faces in frontal-parallel color images. Although most of image processing packages provide functions to beautify color images, a few of them, at least based on our knowledge, are specific for beautifying human faces. By using these functions, the processed face images become unreal. For example, they will remove most of natural edges around some special regions, such as eyes and mouth. Therefore, the proposed system only processed the regions of faces. To make the processed face region smoother, our system treats the regions of eyes/mouth and the rest of face region differently. By using different methods to smooth these two types of regions, we can keep almost all the natural edges around eyes and mouth , but remove wrinkles and spots on the rest of faces. The process of our system ia as follows. At beginning, we convert the RGB color space into YCbCr space so as to segment face region from scene background based on the value range of the skin color proposed by H.A. Rowly, etc. Within the face region, the system uses the chain-code to get the eye region and the mouth region. For the eye and mouth regions, we adjust the image pixels by pixels; the rest of pixels are justified by block base. To evaluate the performance of our system, we compare our system with the tool?cleanSkinFX, which can be found at the Web site http://www.mediachance.com. Our system is outperforming.

Due to the dramatic growth of network and multimedia technology, people can more easily get variant information by using Internet. Unfortunately, it also makes the diffusion of illegal and harmful content much easier. So, it becomes an important topic for the Internet society to protect and safegurard Internet users from these content that may be encountered while surfing on the Net, especially children. Among these content, porno graphs cause more serious harm. Therefore, in this study, we propose an automatic system to detect still colour porno graphs. Starting from this result, we plan to develop an automatic system to search porno graphs or to filter porno graphs. Almost all the porno graphs possess one common characteristic that is the ratio of the size of skin region and-skin region is high. Based on this characteristic, our system first converts the colour space from RGB colour space to HSV colour space so as to segment all the possible skin-colour region from scene background. We also apply the texture analysis on the selected skin-colour region to separate the skin region from non-skin region. Then we try to group the adjacent pixels located in skin region. If the ratio is over a given threshold, we can tell if the given image is a possible porno graph. Based on our experiment, less than 10% of non-porno graphs are classified as pornography, and over 80% of the most harmful porno graphs ara classified correctly.

We have investigated characteristics of a commercially available one-chip white-LED based on indium-doped gallium nitride when modulating with an electrical pulse. The observation of a spectrum of the white-LED revealed that it contained four large peaks at wavelengths of 390nm (ultra-violet), 430nm (blue), 510nm (green), and 620nm (red), respectively. When the LED was modulated by use of a squared pulse signal with a 16kHz repetition frequency and an 85% duty cycle, each color exhibited different electric characteristics: ultra-violet and blue colors showed the similar electrical spectrum of the original input signal having three frequency bands (1st band: 16~108kHz, 2nd band: 108~208kHz, 3rd band: 208~320kHz), meanwhile a green color was presented in the two bands and a red component did not manifested in any bands. The measured result can be explained in terms of different response of the fluorescent materials responsible for the light emission. Use of electrical band filters enables to separate a white light into desired one.

With the strong demand from the imaging industry, the CIE Division 8 Image Technology was established in November 1997. This young and dynamic division is aimed to study procedures and prepare guides and standards for the optical, visual and metrological aspects of the communication, processing, and reproduction of images, using all types of analogue and digital devices, storage media and imaging media. It is a servant for the imaging industry to achieve successful colour practice using the knowledge of colour science and colour engineering. There are six CIE Division 8 Technical Committees (TC): TC 8-01 Colour Appearance Modeling for Colour Management Applications, TC 8-02 Colour Difference Evaluation in Images, TC 8-03 Gamut Mapping, TC 8-04 Adaptation under Mixed Illumination Conditions, TC 8-05 Communication of Colour Information, and TC 8-06 Image Technology Vocabulary. This paper will introduce the aims and activities in each TC.

This new technique for imaging spectrometry is based on computed-tomography (CT). The obvious feature of this method comparing with other imaging spectrometry is its optical system. Two cylindrical lenses compress a two-dimensional object into a one-dimensional image. Subsequently, the spectral image dispersed by a grating is obtained by two-dimensional detector array. A rotation scanning could collect a series of spectral images with different projection angles. The spectral image of the object is reconstructed from these projection images by computed-tomography algorithm. Although this method still needs a rotation scanning to obtain enough information for giving out the data cube of the object, its energy efficiency, and its signal to noise, could be improved significantly. The optical and mathematical theories of this method are discussed. And the suggested optical configuration has also given. A discrete model of this imaging spectrometric system is established to show the digital reconstruction algorithms. A computer simulation based on the discrete model is executed by MATLAB. The reconstructed results with a defined spectral image indicate that the method is capable of reconstructing scenes containing both broadband samples and narrow-band samples. When the specific noise is added, the impact of the number of projection angles on reconstruction precision is analyzed.

A high-speed color measurement system is structured with high precision performance. It is addressed the principle of color measurement and the structural characters of the system. The hardware and software arrangement are also included. This system is equipped with the linear Charge-Coupled Device (CCD) as the photo detector, the pulsed-xenon lamp as the light source, and concave holographic diffraction grating as a polychromator. As the result of such the photoelectric arrangement, the measurement accuracy is improved a lot and the repeatability is also very satisfactory.

Neugebauer equations are basic formulas to calculate colors of print. But there is obvious deviation between calculation and measurement due to the complexity of printing process. Some methods of improving its precision are reviewed and an improved spectral Neugebauer method is proposed in this paper, which is based on spectral dot gain correction. There are two kinds of dot gains, one is mechanical dot gain and another is optical dot gain. Both of them are the key factors affecting calculated results of Neugebauer equations. It is found that the optical dot gain is a function of wavelength, so that the correction should also be a function of wavelength. After measuring spectral curves of paper substrate, primaries and their mixture, spectral dot gain can be calculated and Neugebauer equations can be corrected with it. The forwards color lookup table is got directly from the improved Neugebauer equations, the reverse color lookup table is calculated by interpolation from forwards color lookup table (LUT).

There are lots of experimental results concerned with chromatic contrast detection (based on changes in chromatic distribution over space or time), but compared with luminance contrast detection, there has been no theory to describe chromatic contrast detection and thus no mathematic model for isoluminant contrast sensitivity function. After reviewing of many papers, two reasons may contribute to this: I . Isoluminant contrast sensitivity function has the same description method as the isochromatic contrast sensitivity function which is unfit to express the chromatic changes. 2. The color vision is so complicated that further study is needed. Based on detailed analysis of previous works, this paper utilizes colorimetry fundamental principle to present a new theory for color contrast detection, and indicates that isoluminant contrast sensitivity function can be described by two aspects of chromatic changes——dominant wavelength and colorimetric purity. This theory can well explain that red-green and yellow-blue contrast sensitivity functions have similar characteristics.

Use of electronic video endoscope for viewing the lumen by clinicians is frequently carried out. Computer-assisted for processing and analysis the images will aid non-expert clinicians in diagnosis. Abnormalities can be detected by characterizing the features of the segmented endoscopic images. A new technique has been developed for segmenting the endoscopic image obtained from the large intestine using color parameters for detecting the lumen. This novel method for endoscopic image segmentation exploits the local homogeneity index (?) definition of the pixels in a given neighborhood of the image. The proposed segmentation is realized in two stages. Initially, segmentation is performed using ?, which is measured in terms of standard deviation and discontinuity in the achromatic distribution of the region. A modified peak-finding algorithm is employed to segment the image from the corresponding histogram. In the second stage, the regions obtained in the first stage are divided into subregions based on ? in the chromatic domain. Merging the region using color difference measures alleviates the problem of oversegmentation. This novel method is compared with a minimum variance region-growing method. The proposed technique is tested using clinically obtained colonoscopic images. The preliminary results indicated that both techniques are feasible. The technique based on ? showed better performance than a minimum variance region-growing method because it includes the analysis of both local and global information by using ? histogram. The proposed scheme can be adopted for computer-based analysis of the endoscopic information to facilitate early detection of colorectal cancer.

Dense yttria stabilized zirconia(YSZ) coatings have been prepared by the sol-gel method with zirconium oxychioride as a precursor. By controlling the processes and conditions, the coatings on the porous zirconia substrate are crack-free, have appropriate thickness and e. When exposing the surface of zirconia coatings to Lasers, changes of contact angles between water and the exposed surface are found. The crystalline structure of yttria stabilized zirconia are identified by X-ray diffraction (XRD); the morphology of coatings are observed by scanning electron microscope (SEM); the contact angels between surfaces and oil (or water) are measured by contact angel measuring system, It is shown that the contact angels of exposed surface of YSZ coatings increase significantly.

This paper introduces a novel methodology for object detection using genetic algorithms and morphological processing. The method employs a kind of object oriented structuring element, which are derived using genetic algorithm operating. The population of morphological filters iteratively evaluated according to a statistical performance index corresponding to object extraction ability, and evolves into an optimal structuring elements using the evolution principles of genetic search. Experimental results of object extraction in high resolution satellite images are presented to illustrate the merit and feasibility of the proposed method.

This paper describes a fast and effective approach for fingerprint image preprocessing. It is suitable for the minutiae matching because the approach could filter out error skeletons in the fingerprint image. Traditional preprocessing of fingerprint recognition uses quantization interval to adjust the pixel values in a gray-level fingerprint image to clear the fingerprint ridges. Then, it applies eight direction windows to modify the fault direction of the fingerprint ridges. The third stage is converting the gray-level fingerprint image to black-while one, and thins the fingerprint ridges to their skeletons with one-pixel depth. However, the above procedures may also result error skeletons, such as branches, noises, and gaps. We therefore need to filter those errors out to raise the recognition rate. Experimental results show that our approach is not only simple and fast, but also has the ability to delete all kinds of error skeletons. Hence the approach suggested in the paper should be appropriate for the minutiae matching.

Photoelectron plays an important role in latent image formation in silver halide materials, and its decay process is connected with the photographic process. Therefore, more details about the photoelectron decay behavior are needed in order to improve the photographic efficiency of imaging materials. In this paper, we use a single computer to simulate the photoelectron decay process in AgC1 microcrystals doped with [Fe(CN)6]4- that acts as shallow electron traps (SETs). First, we propose a model, which consists of a [Fe(CN)6]4- related SETs and intrinsic centers of pure AgC1 including two types of electron traps and a recombination center, where first a hole is trapped that then combines with a free photoelectron. The model results in a set of differential equations that describe the kinetics of generation, trapping, thermal detrapping and recombination processes. The decay process of the photoelectrons is simulated through solving these differential equations. In this simulation, the photoelectron decay curves, which fit well with experimental results, can be obtained quickly and accurately. With the aid of the simulation, the decay curves of the free photoelectron and the shallow-trapped electron are obtained accurately. A number of important conclusions about the SETs were drawn from the simulation study.

Over last two decades, 3-D vision methods have been developed rapidly and have become an important research field. Structure light 3-D vision technology is a kind of active 3-D vision method. When the structure light's pattern are coded by black and white stripes, we are prone to get incorrect data because of difficulty of seeking correspondence relation, therefore conventional 3-D vision technology using structure light has some limits. In this paper, we put forward a new structure light pattern based on hue information. Each color in real world can be signified by hue, saturation and lightness, and we designed a kind of new pattern, which includes eighteen colors. The shape of objects was acquired and reconstructed though triangular relation. According to difference of hue, processing software can find corresponding relation between distorted stripes and originally regular stripes, and avoid incorrect corresponding using conventional pattern of structure light, such as B/W stripes. Finally the experiment result was given.

In the previous literature, the focusing measures usually analysis the gray image converted from color image when the focusing target image is color one. Here a fast auto-focusing approach based on green components analysis is put forward. The color CCD (Charge Coupled Device) has three types elements. They are respectively sensitive to red light (so be called Red element), green light (Green element), or blue light (Blue element). But they are not very "pure". For example, the green element is not only sensitive to green light, but also sensitive to red one and blue one. The most photography scenes always include the "Green object" which is sensitive to the above green element. So it is possible to analysis only the green image instead the gray one converted from color one, to decide the taken image is focused or not. The results show the fast auto-focusing approach based on green image analysis is similarly valid. Because the time which be used to get gray image from color image in general auto-focusing approach has been omitted, the auto-focusing speed of digital system will be increased about 2 times by using our new approach. In the paper, the auto-focusing time has been discussed.

In this paper, we report a novel method of fuzzy color-image retrieval based on color histogram correlation, for which peaks of the histograms are first ranked according to their heights and then the correlation based on both height and position will be calculated for correspondent rank respectively. A correlation result will be defined by ?- cut procedures. Experiment results confirmed the effectiveness of the method.

A color-encoded technique for 3D measurement is presented in this paper. In this technique, a color pattern that is composed of RGB components is created by software on a computer screen and projected to an object by a digital light processing, which is controlled by the computer. The image of the object is captured by a CCD color digital camera positioned at an angle different from that of the digital projection system. The color of each pixel of the image has a one-to-one correspondence with the projection angle. The 3D depth map of the object is obtained by way of computer decoding and calculating of the color image captured. The 3D measuring system and its parameters are designed based on the principle of color-encoded technology.

Photo-electron decay characteristics in imaging process directly decide photographic efficiency of AgX emulsion. Microwave Absorption and Dielectric Spectrum Measure Technique (MADSMT) is a powerful tool to detect the change of emulsion film illuminated by light Through detecting the changes of reflect voltage in MADSMT, the decay signals of free electrons and shallow-trapped electrons will be obtained. In our work, the sulfur-sensitized silver halide material samples are used, which are made in different conditions, for example different temperatures and different densities. In order to record the whole decay process, a YAG laser system whose pulse width is 35ps is used as an exposure light. Finally the mechanism of sulfur-sensitized centers has been discussed and the influence of different conditions has been compared and analyzed.

The relationship between input DAC count and output luminance in CRT display varies according to the “brightness” control. In this study the effect of brightness adjustment on the accuracy of color reproduction in CRT display is shown experimentally. The optimum brightness level of test monitor has been measured by using a device developed in our laboratory. The accuracy of the simplified color reproduction model is evaluated for three monitor setups, which is controlled to optimum, low, and high brightness level. To prove the experimental results, the channel additivity and chromaticity constancy assumptions are tested for three setups. With the color difference ?E*u'v' between predicted and measured values for 215 colors, it can be ascertained that the brightness must be adjusted to optimum level to enhance the accuracy.

A new method for the car plate extraction and recognition is presented in this paper. First, the segmentation technique, based on multiple threshold values determined by fuzzy entropy and the prior knowledge on car plate character, is used to extract the car plates, which has strong anti-noise capability, and is able to locate the car plates quickly in the varying backgrounds. Then, a recognition method combining the several recognizers is proposed to recognize the car plate characters. The recognition rates of over 97% under various illumination conditions in real applications shows that the proposed method is effective and reliable for car plate recognition.

In cross-media color image reproduction, gamut mapping is needed due to gamut difference among different media. In order to implement gamut mapping, gamut boundaries of each medium involved should be first determined. It may be expected that an analytical expression for a boundary is preferred than a set of discrete data , since it would take less storage space and make the determination of the intersection point between a boundary and a "mapping line" easier and faster. In this article, a form of Zernike polynomial expression is suggested to be used as the expression of gamut boundary surface. For instance, if C1E1976L*a*b* is adopted as the color space for gamut mapping, then each color(point) on the boundary can be expressed as L*=L*(a*,b) and the boundary can be expanded into a series of Zernike polynomials with an appropriate coefficient for each of which. These coefficients can be obtained with sufficient experimental data of boundary points and existing algorithms. Experiments have been executed for a color printer with(R,G,B) as its input. The 6 boundaries in RGB space would consist of (0,G,B),(R,0,B),(R,G,0),(255,G,B),(R,255,B) and (R,G,255) where each of R,G,B varies from 0 to 255. Then 6 corresponding sets of Zernike coefficients are calculated, based on about half of the measured L*a*b*'s for each boundary. A comparison between original measured data and the data predicted by Zernike polynomials shows that, not only for the data that have been used to calculate the coefficients, but also for those not used, the differences are acceptably small even negligible with only a few exceptions.

For the training of the BP neural networks in CRT color conversion, some papers suggest using a uniformly distributed RGB training set model (URGB). However, this URGB model is single-directional. Therefore, when the number of the samples in a training set is under a certain amount, such as less than 51 2 (8 X 8 X 8), a URGB model may cause big prediction errors, especially in the backward conversion (XYZ to RGB). In this paper, we propose an improved training set model, with which a smaller training set can be drawn from a virtual URGB set. Our experimental results show that, an improved training set model can achieve a desired prediction accuracy in the whole CRT color space, even if the samples number in a training set is less than 512(8 X 8 X 8).

A psychophysical experiment was carried out with the method of constant stimuli using CRT-generated color samples. The experimental results at the five CIE color centers of Gray, Red, Yellow, Green, and Blue were satisfactorily described by chromaticity ellipses as equal color-difference contours in the CIELAB space. The CIEDE2000 formula, together with other two advanced color-difference formulae, CMC and CIE94, and the basic CIELAB equation, were tested using the visual data obtained from the present experiment. The comparisons between color differences, ?E, predicted by individual formulae and the corresponding visual scales, ?V, were carried out in terms of PF/3 measure. With their original forms, i.e. kL=kc=kH=l, or with their optimized kL values, the CIEDE2000 outperformed others for the combined dataset under the viewing condition in this study. Furthermore, the visual data at Blue center were well predicted by CIEDE2000 with an obviously better accuracy than other color-difference formulae. This confirms that the rotation item involved in the CIEDE2000 equation effectively improves the uniformity and predicting performance for the color differences in the blue region.

In this paper, a new whiteness formula in the CIELUV uniform color space is developed. The results show that it is superior to the CIE whiteness formula and the others in visual correlativity, uniformity and applicability.

As new method of characterizing CRT monitors is proposed. The features of this method are, it can take account of some color appearance factors, such as the appearance difference between self-luminous and surface color, but without the complexity of using any color appearance model, and it may improve the performances of an interpolation operation when an arbitrary assigned color is to be displayed on a CRT screen. The method is introduced more detailedly in Section 2, and preliminary experimental results are given in Section 3.

It is necessary for evaluating the color reproduction from CRT image to printer image. However, the process of evaluation is difficult. This paper describes an approximate, but very simple and objective method of evaluating the color reproduction from CRT to printer by employing a scanner. The process is as follows: (1)an experimental image on a CRT, witch contains 1301 color patches with size 6x6mm2 ;(2)the printed image was scanned by a scanner; (3)segmenting the scanned image for finding each color patch, then calculating average values of RGB for each patch; (4)calculating color difference between scanned image and original CRT image for each patch. Average difference may show approximately the color difference between a CRT image and the printed image. This method is based on a fact: color reproduction from CRT to printer is worse than from printed image to the scanned image. In our experiments, average CIE 1976Lu*v* color difference between the printed image and the CRT image is 10.9 units, but average color difference between the scanned image and the printed image is less than 4 units. Experiments show that the result of evaluation by this method is consistent with one by observers visually.

For some food, the color shows the character and it is an important criterion to determine the fresh of food. This paper presents an optical method to measure the food color for determining the fresh level, the processing quality and distinguishing the true from the false of the food and so on. The reflection spectrum is used to measure the surface color of solid food and the principle to deal with the color in digital is discussed. As an example, the fresh of meat is change along the inductor of meat chromo protein, but it is difficult to measure the inductor because it exists as mixed along with the time. In this paper, the digital model between the hue and the inductor of meat's chromo protein is made, and then the model appraises the fresh of meat. The feasibility of the method is proved by the test using relationship of the meat's color and its fresh.

In general, the high precise color meter uses spectrophotometer to measure sample's reflection or transmission rate and then calculate all kinds of color parameters. About spectrophotometer, to design the monochrome parts is very important. At present, the spectrophotometer designed for color meter never use grating scanning but linear array sensors. Its advantage is greatly reducing the measuring time and making the miniature and portable instruments possible. This article will select the linear photodiode array as an example to discuss how the relationship among the wavelength ?, the grating parameter d, the incidence angle ?, the focus length f', and the number of sensitive cells n in the array. We will deduce the formula of them and calculate the theory results. At last we analyze the difference between the real results and the theory results, discuss some reasons and recommend some methods to improve the wavelength accuracy. These formula and methods will be very useful for spectrophotometer design based on linear array sensors.

In the present work we have designed an small-size portable imaging spectrometer using linear variable interference filter, with CCD imaging head as picture sensor, and micromove stage to bring along interference filter which scan across every pixel of CCD array. The difference column pixel in the picture which we get at every step is monocolour partial image under different wavelength, with the step by step, these column monocolour pixels change the wavelength. And finally we reassembled these images and get the whole monocolour image with different wavelength. The interval of sweep step length decided by the required spectrum resolution and the required wavelength interval of different pictures. The experiment demonstrated the resolution of this spectrometer is about l6nm. But that major reason of the limiting resolution is the band pass of the linear variable interference filter. The spatial resolution ofthe instrument ultimately decided by CCD and imaging lens. This spectrometer has some characters such as compact structure, higher spatial resolution, higher spectrum resolution, higher scan rate and so on.

The process of color image reproduction from scanners to printers includes a series of steps. For realizing color reproduction, artificial neutral networks are used to each of steps separately or the whole process. However, In applications, these technologies are still complex. For convenience to applications, this paper presents a simple method of color image reproduction from scanner to printer and color prediction before printed using a back-propagation neutral network, which maps RGB values between scanner and printer directly and teacher values can be gotten automatically. The processes of color production and color prediction use the same network structures, same leaming data and inverse learning method. Experimental results show that, using this method of color reproduction average color difference ?Eab between the print-out image and scan-in image are reduced from 15 to 4 units. The network has better behave for color prediction than color reproduction, and the reason is discussed.

A uniform colour space (UCS) would be very useful for image applications such as gamut mapping, image compression, and evaluation of colour management and colour reproduction systems. Current colour spaces, however, can only be applied under a fixed set of viewing conditions. It is essential to construct a uniform colour space, which can predict colour difference under a wide range of viewing conditions. This paper describes a new uniform colour space J 'a 'b 'actually based upon one of the revised colour appearance models, CAM97s2 to fit the available small colour difference datasets including BFD-P, RIT-DuPont , Leeds and Witt . Our results show that J 'a 'b ' fits the above experimental datasets better than the current CIELAB and DIN99 spaces.