In this article, we propose a new class of multichannel filters, vector rank M-type K-nearest neighbor (VRMKNNF) filters that can obtain the balance between noise suppression, and edge and fine detail preservation, especially in color image restoration. The proposed VRMKNNF filters are mainly based on the combined RM-estimators with different influence functions. An adaptive non parametric approach that determines the functional form of the density probability of noise from data into the sliding filtering window has been employed to improve the performance of the multichannel filters. Applying the VRMKNNF as a reference filter we designed the adaptive multichannel non parametric VRMKNNF (AMN-VRMKNNF). Numerous simulations presented in the article illustrate that the VRMKNNF and AMN-VRMKNNF filters exhibit the robust and adaptive capability in multichannel imaging applications. Finally, we present the implementation of proposed filters on the DSP TMS320C6711 demonstrating that they can potentially provide a real-time solution to quality video transmission.

This article proposes a real-time method for the detection and elimination of specular reflectance in color images. We use a 2D histogram that allows us to relate the signals of luminance and saturation of a color image and to identify the specularities in a given area of the histogram. This is known as the LS diagram and it is constructed from the HLS color space. A detailed study of the presence of specularities in the diagram for different types of materials is carried out. To eliminate the specularities detected, we use a new connected vectorial filter based on color morphology and adapted to real-time specifications. This filter operates only in the bright zones previously detected, reducing the high cost of processing of connected filters and avoiding oversimplification, in single processing and multiprocessing environments. The new proposed method achieves good and similar results to the ones obtained with other techniques used in multimedia, but it not requires costly multiple-view systems or stereo images.

We describe a novel real-time image processing and sensor fusion system for aerial vehicles in need of autonomous landing, guidance and obstacle avoidance. The Intelligent Synthetic Vision System (ISVS) described herein can process and display information merged from multiple image sources including a high resolution millimeter wave radar, a stored terrain with 3D airport database, FLIR, as well as visual reference images. The resulting solution provides safe all-time/all-visibility navigation in terrain-challenging areas and, in particular, real-time object detection during the final phases of landing in all weather conditions.

Real-time three-dimensional (3D) medical imaging requires both high memory and high computational bandwidth due to the massive amount of data to be processed. Application-specific systems have been designed to speed up a few selected 3D imaging algorithms. When applied to other algorithms, all such systems require complicated redesign procedures or get lower performance than expected. In this article, we present an FPGA-based computing platform that can be used to accelerate a broad range of 3D medical imaging algorithms dominated by local operations. Its generality and reconfigurability make it easy to be customized to differing algorithm requirements. The architecture in the platform exploits intrinsic parallelisms in 3D imaging algorithms in order to achieve high computational bandwidth. Along with the architecture, a new caching scheme called brick caching was designed to dynamically partition the input data and buffer them in distributed internal caches to provide multiple memory accesses. The caching scheme exploits locality of reference in three dimensions, thus greatly reducing the total data flow from external system memory to internal caches. It is also a deterministic caching method, which allows the input data to be prefetched into the processor before they are processed. We also present application of the platform in FDK (Feldkamp–Davis–Kress) cone-beam computed tomography (CT) image reconstruction so as to demonstrate its potential in real-time 3D medical imaging.

The aim of this study is to develop a color visualization system for multispectral images in the near-infrared region (NIR, 800–1000 nm). Samples with the same color or appearance in the visible region, which are therefore indistinguishable to the human eye, can have different reflectance or transmittance spectra in other parts of the electromagnetic spectrum, specifically in the NIR. Therefore, these samples can be differentiated by taking into account this extra information. In this work, we use a multispectral system that we have developed recently in order to obtain five images of several samples with varying spectral reflectance, corresponding to different spectral bands. We then define a color space representation which associates the camera responses to the color channels of a calibrated CRT monitor. Therefore, a pseudo-colored image is obtained. Several possible associations are presented, some of them based on methods which attempt to imitate human color vision but in the NIR region, and others for maximizing colorimetric discrimination between the objects, based on principal component analysis (PCA). Finally, the color differences between samples are evaluated using several parameters. The methods which provide the best results in terms of visual discrimination are based on PCA analysis, but the methods related to human color vision keep information of the NIR spectral reflectance of the samples.

This article reports on the colorimetric improvement of a multispectral 3D digitizer through scanning at a small number of nearoptimal wavelengths. Optimal wavelengths were first investigated theoretically based on the criterion of minimal color differences over some sets of reflectance curves using PCA-based and spline-based spectral estimation methods. Sets of three, four and five optimal sampling wavelengths were derived to within ±5 nm for each one based on the spread of variation for different combinations of the method and the data set used for the derivation. This provided a basis for the selection of HeCd, ArKr, HeNe and DPSS commercial laser lines for which the colorimetric performance was predicted. This was then tested in the laboratory, where color rendition charts were scanned with the camera at seven wavelengths, after which the charts were computer rendered on a CRT display. Both the theoretical prediction and the experimental observation indicate that four well-chosen wavelengths are adequate for proper rendition of the charts.

Chromatic induction effects depend, among others, on the frequency content of the observed region. Highly textured images usually show some dominant frequency information, which produces prominent chromatic induction effects when observed. As it is shown by some authors, the two chromatic induction effects, i.e., chromatic contrast and assimilation, can be computationally simulated by blurring and sharpening operators, respectively. In this article, we present a first approximation to the perceptual representation of highly textured images using a wavelet decomposition approach. Wavelet coefficients are modulated by a weighting function, which performs either assimilation or contrast at every frequency level of the image. This wavelet approach allows us to define both chromatic induction effects in a unified framework as a single mathematical operator.

In this study the purpose was to develop a computational surface model, which gives an equal computed chromaticity difference for equal perceived chromaticity difference in every part of the color space. The computational surface model assess chromaticity differences based on the surfaces defined by an ellipse data set and the two chromaticity points whose chromaticity difference is to be calculated. The distances along the surfaces are calculated by a method based on the Weighted Distance Transform On Curved Space (WDTOCS). The ellipse data sets are the MacAdam ellipses in the CIE 1931 (x,y)-chromaticity diagram and the ellipses which are fitted from the visual color difference measurements used in deriving the CIE DE2000 color difference formula in the CIELAB color space. In general the ellipse data set can be any set of planar chromaticity ellipses. The chromaticity differences calculated along the surfaces correct the planar chromaticity values, thus these differences match better with perceived chromaticity differences. The experiments are made in the vicinity of the chromaticity difference ellipses and the results are contrasted with CIE DE2000.

Optical properties of handsheets made from pulp resulting from deinking flotation of multicolored and desaturated prints from digital and conventional offset printing have been observed. The prints were made on different models of digital offset machines. One series of prints was naturally aged for the period of one year and the other was accelerated aged in a climatic chamber. The results show that the model of the printing machine, even with the same technology, can affect the size of particles and the optical properties of the handsheet. The quality of deinking pulp and the efficacy of the ink removal in relation to the optical parameters such as brightness, reflectance in the determined areas of wavelengths and color are discussed in this work.

From a human centered perspective three ingredients for Content-Based Image Retrieval (CBIR) were developed. First, with their existence confirmed by experimental data, 11 color categories were utilized for CBIR and used as input for a new color space segmentation technique. The complete HSI color space was divided into 11 segments (or bins), resulting in a unique CBIR 11 color quantization scheme. Second, a new weighted similarity function was introduced. It exploits within bin statistics, describing the distribution of color within a bin. Third, a new CBIR benchmark was successfully used to evaluate both new techniques. Based on the 4050 queries judged by the users, the 11 bin color quantization proved to be useful for CBIR purposes. Moreover, the new weighted similarity function significantly improved retrieval performance, according to the users.