Visual discomfort has been the subject of considerable research in relation to stereoscopic and autostereoscopic displays. In this paper, the importance of various causes and aspects of visual discomfort is clarified. When disparity values do not surpass a limit of 1°, which still provides sufficient range to allow satisfactory depth perception in stereoscopic television, classical determinants such as excessive binocular parallax and accommodation-vergence conflict appear to be of minor importance. Visual discomfort, however, may still occur within this limit and we believe the following factors to be the most pertinent in contributing to this: (1) temporally changing demand of accommodation-vergence linkage, e.g., by fast motion in depth; (2) three-dimensional artifacts resulting from insufficient depth information in the incoming data signal yielding spatial and temporal inconsistencies; and (3) unnatural blur. In order to adequately characterize and understand visual discomfort, multiple types of measurements, both objective and subjective, are required.
In this paper, we present an intensity mapping algorithm for enhancement and segmentation of pancreatic ductal nuclei in pancreatic ductal cell images acquired from pancreatic fine needle aspiration specimens with Papanicolaou stain. The upper envelope surface is obtained by rolling a ball of fixed size on the image surface followed by a moving average filtering to smooth out the downward spikes. When the upper envelope surface is mapped to the uniform intensity of the maximum intensity of the original image, the contrasts inside nuclei are magnified and the difference between intensities in the cytoplasm regions and the void background is compressed, resulting in improved separation of the nuclei from the rest. Results of enhancement and segmentation of pancreatic nucleus images are provided and compared to those without envelope mapping.
The Yule–Nielsen modified spectral Neugebauer model (YNSN) enhanced for accounting for ink spreading in the different ink superposition conditions provides accurate spectral predictions, but requires one to measure the reflectances of special halftone calibration patches in order to compute the ink spreading curves mapping nominal ink surface coverage to effective ink surface coverage. Printing special halftone calibration patches within the borders of the print pages is cumbersome, since these special patches need to be cut out before assembling the final printed product. In the present contribution, we calibrate the ink spreading curves directly from the printed images by fitting them so as to minimize a distance metric between predicted reflectances and measured reflectances at selected relatively uniform locations. We compare the prediction accuracy of the so-calibrated YNSN model with the prediction accuracy of the same model calibrated by the spectral reflectances of classical uniform calibration patches. Interestingly, when the model is calibrated from image tiles originating from the same or from a similar color image as the one comprising the test tiles, better prediction results are obtained than when performing a classical calibration on 50% halftone patches printed in all superposition conditions.
This paper proposes a driving current control method for a back light unit (BLU), consisting of red, green, and blue (RGB) light-emitting diodes (LEDs), whereby a RGB optical sensor is used to check the output color stimulus variation to enable a time-stable color stimulus for light emission by the RGB LED BLU. First, to obtain the present color stimulus information of the RGB LED BLU, an RGB to XYZ transform matrix is derived to enable CIEXYZ values to be calculated for the RGB LED BLU from the output values of a RGB optical sensor. The elements of the RGB to XYZ transform matrix are polynomial coefficients resulting from a polynomial regression. Next, to obtain the proper duty control values for the current supplied to the RGB LEDs, an XYZ to Duty transform matrix is derived to calculate the duty control values for the RGB LEDs from the target CIEXYZ values. The data used to derive the XYZ to Duty transform matrix are the CIEXYZ values for the RGB LED BLU estimated from the output values of the RGB optical sensor and corresponding duty control values applied to the RGB LEDs for the present, first preceding, and second preceding sequential check points. With every fixed-interval check of the color stimulus of the RGB LED BLU, the XYZ to Duty transform matrix changes adaptively according to the present lighting condition of the RGB LED BLU, thereby allowing the RGB LED BLU to emit the target color stimulus in a time-stable format, regardless of changes in the lighting condition of the RGB LEDs.
Today, CCD and CMOS detector arrays present excellent features in imaging systems. To investigate the suitability of each technology according to various applications, in this work we have comparatively studied the quality of the images provided by different cameras. To this end we have used the speckle method to determine the modulation transfer function (MTF) at different wavelengths of the visible spectrum for the detectors of a low-cost CCD video camera and of two scientific cameras (CCD and CMOS). For both the low-cost CCD and the scientific CMOS detectors, the differences between the MTF curves intensify as the spatial frequency augments, while the MTF decreases as the wavelength increases. For the scientific CCD detector, the MTF spectral behavior does not show this trend, and the differences between the MTF curves corresponding to extreme wavelengths are not expected to be significant, as opposed to what appears for the scientific CMOS detector.
We describe a method to calibrate the elements of a multispectral imaging system aimed at skylight imaging, which consists of a monochrome charge-coupled device (CCD) camera and a liquid crystal tunable filter. We demonstrate how to calibrate these two devices in order to build a multispectral camera that can obtain spectroradiometric measurements of skylight. Spectral characterizations of the tunable filter and the camera are presented together with a complete study of correcting temporal and spatial noise, which is of key importance in CCDs. We describe all the necessary steps to undertake this work and all the additional instrumentation that must be used to calibrate the radiometric devices correctly. We show how this complete study of our multispectral system allows us to use it as an accurate, high resolution spectroradiometer.
We set up a multispectral image acquisition system using a flash light source and a camera featuring optical bandpass filters. The filters are mounted on a computer-controlled filter wheel between the lens and a grayscale sensor. For each filter wheel position, we fire the flash once within the exposure interval and acquire a grayscale image. Finally, all grayscale images are combined into a multispectral image. The use of narrowband filters to divide the electromagnetic spectrum into several passbands drastically reduces the available light at the sensor. In case of continuous light sources, this requires powerful lamps producing a lot of heat and long exposure times. In contrast, a flashgun emits its energy in a very short time interval, allowing for short exposure times and low heat production. Our detailed colorimetric analysis comparing the color accuracy obtainable with our flashgun and a halogen bulb shows that our acquisition system is well suited for multispectral image acquisition. We computed a mean color error of 1.75 CIE ΔE00 using the flashgun. Furthermore, we discuss several practical aspects arising from the use of flash light sources; namely, the spectrum, repeat accuracy, illumination uniformity, synchronization, and calibration of the system. To compensate for intensity variations of the flash, we propose two calibration methods and compare their performance.
Modeling image naturalness has been a very challenging topic. The ideas in this work represent a new approach, since earlier studies considered only memory colors. Perceived naturalness was assessed for eight test images and their 22 manipulations. These were rendered in the lightness, chroma, or sharpness domain using a nine-point qualitative category scale. From the results of psychophysical experiments, four visual phenomena were determined to be important factors in the judgment of naturalness. These factors were modeled using three components–image sharpness, image colorfulness, and reproduction of shadow detail (and the absence of a washed-out appearance)–in an image naturalness model. Correlates of the three components were derived using parameters optimized from color-appearance attributes and pixel-based color difference in CIECAM02 and CAM02-UCS space. It was demonstrated that the developed image naturalness model could predict well the perceived naturalness changes arising from images varying in the lightness, chroma or sharpness domains. Additionally, it was also argued that the application of memory colors to modeling naturalness is not sufficient.
The estimation of the illuminant of a scene from a digital image has been the goal of a large amount of research in computer vision. Color constancy algorithms have dealt with this problem by defining different heuristics to select a unique solution from within the feasible set. The performance of these algorithms has shown that there is still a long way to go to globally solve this problem as a preliminary step in computer vision. In general, performance evaluation has been done by comparing the angular error between the estimated chromaticity and the chromaticity of a canonical illuminant, which is highly dependent on the image dataset. Recently, some workers have used high-level constraints to estimate illuminants; in this case selection is based on increasing the performance on the subsequent steps of the systems. In this paper the authors propose a new performance measure, the perceptual angular error. It evaluates the performance of a color constancy algorithm according to the perceptual preferences of humans, or naturalness (instead of the actual optimal solution) and is independent of the visual task. We show the results of a new psychophysical experiment comparing solutions from three different color constancy algorithms. Our results show that in more than half of the judgments the preferred solution is not the one closest to the optimal solution. Our experiments were performed on a new dataset of images acquired with a calibrated camera with an attached neutral gray sphere, which better copes with the illuminant variations of the scene.
In this paper, we present a fuzzy-set of parametric functions, which segment the CIELAB space into 11 regions, which correspond to the group of common universal categories present in all evolved languages as identified by anthropologists and linguists. The set of functions is intended to model a color-name assignment task by humans and differs from other models in its emphasis on the intercolor boundary regions, which were explicitly measured by means of a psychophysics experiment. In our particular implementation, the CIELAB space was segmented into 11 color categories using a triple-sigmoid function as the fuzzy-sets basis, whose parameters are included in this paper. The model's parameters were adjusted according to the psychophysical results of a yes/no discrimination paradigm where observers had to choose (English) names for isoluminant colors belonging to regions in between neighboring categories. These colors were presented on a calibrated CRT monitor (14-bit×3 precision). The experimental results show that intercolor boundary regions are much less defined than expected, and color samples other than those near the most representatives are needed to define the position and shape of boundaries between categories.