This paper addresses the problem of determining the 3-D pose of a curved rigid object from a single 2-D image. The surface of the object are assumed to be modeled with several patches, each of which be expressed by an implicit polynomial. Moreover, the sensed data are assumed to be the coordinates of those points that are on the image contours. Based on the idea of contour matching, the algorithm proposed computes the parameters defining the pose of the object, and achieves the segmentation of the sensed data and the recognition of the object.

A method for reconstructing realistic 3D human faces from computer tomography images and color photographs is proposed in this paper. This can be linked easily with the underlying bone and soft tissue models. An iteration algorithm has been developed for automatically estimating the virtual camera parameters to match the projected 3D CT image with 2D color photographs using known point correspondence. An approach has been proposed to select landmarks using a mouse with minimum error. Six landmarks from each image have been selected for front face matching and five for each side face matching.

It is urgently required to protect copyrights of digital contents since the digital contents can be easily copied without degradation of quality. In this paper, we propose a new watermarking method which spreads an ID pattern with a random sequence and embeds it throughout the spatial domain of a color image. The random sequence is a key for extracting the ID pattern. As an ID pattern is spread throughout an image, we can extract the ID pattern from a part of the image, that is clipped image. We can also confirm authenticity by extracting the same ID pattern from several parts of an image. The proposed method is robust to disturbance by noise addition and image conversion such as brightness-contrast conversion and JPEG compression.

This paper describes methods of tracking of moving objects in a cluttered background by integrating optical flow, depth data, and/or uniform brightness regions. First, a basic method is introduced which extracts a region with uniform optical flow as the target region. Then an extended method is described in which optical flow and depth are fused. A target region is extracted by Baysian inference in term of optical flow, depth and the predicted target location. This method works only for textured objects because optical flow or depth are extracted for textured objects. In order to solve this problem, uniform regions in addition to the optical flow are used for tracking. Realtime human tracking is realized for real image sequences by using a real time processor with multiple DSPs.

In this paper we propose a method to aid a visually impaired person in the operation of a computer running a graphical user interface (GUI). It is based on image processing techniques, using images taken by a color camera placed over a Braille display. The shape of the user's hand is extracted from the image by analyzing the hue and saturation histograms. The orientation of the hand, given by an angle θ with the vertical axis, is calculated based on central moments. The image of the hand is then rotated to a normalized position. The number of pixels in each column of the normalized image is counted, and the result is put in a histogram. By analyzing the coefficient of asymmetry of this histogram, it can be determined whether the thumb is positioned along the pointing finger, or whether it is far from the other fingers. These two positions define two states that correspond to a mouse button up or down. In this way, by rotating the hand and moving the thumb, we can emulate the acts of moving a scroll bar and depressing a mouse button, respectively. These operations can be used to perform tasks in a GUI, such as cut-and-paste, for example. Experimental results show that this method is fast and efficient for the proposed application.

In this paper, we propose a new approach for describing image patterns. We integrate the concepts of multiscale image analysis, aura matrix (Gibbs random fields and cooccurrences related statistical model of texture analysis) to define image features, and to obtain the features having robustness with illumination variations and shading effects, we analyse images based on the Topographic Structure described by the Surface-Shape Operator, which describe gray-level image patterns in terms of 3D shapes instead of intensity values. Then, we illustrate usefulness of the proposed features with texture classifications. Results show that the proposed features extracted from multiscale images work much better than those from a single scale image, and confirm that the proposed features have robustness with illumination and shading variations. By comparisons with the MRSAR (Multiresolution Simultaneous Autoregressive) features using Mahalanobis distance and Euclidean distance, the proposed multiscale features give better performances for classifying the entire Brodatz textures: 112 categories, 2016 samples having various brightness in each category.

In this paper, we give a solution to the problem of conflict-free access of various slices of data in parallel processor for image processing. Image processing operations require a memory system that permits parallel and conflict-free access of rows, columns, forward diagonals, backward diagonals, and blocks of two-dimensional image array for an arbitrary location. Linear skewing schemes are useful methods for those requirements, but these schemes require complex Euclidean division by prime number. On the contrary, nonlinear skewing schemes such as XOR-schemes have more advantages than the linear ones in address generation, but these schemes allow conflict-free access of some array slices in restricted region. In this paper, we propose a new XOR-scheme which allows conflict-free access of arbitrarily located various slices of data for image processing, with a two-fold the number of memory modules than that of processing elements. Further, we propose an efficient data alignment network which consists of log N + 2-stage multistage interconnection network utilizing Omega network.

It becomes essential in practice to improve a processing rate and to divide an image into small segments adjusting a limited memory, because image filing systems handle large images up to A1 size. This paper proposes a new method of an automatic skew normalization, comprising a high-speed skew detection and a distortion-free dividing rotation. We have evaluated the proposed method from the viewpoints of the processing rate and the accuracy for typed documents. As results, the processing rate is 2. 9 times faster than that of a conventional method. A practical processing rate for A1 size documents can be achieved under the condition that the accuracy of a normalized angle is controlled within 0. 3 degrees. Especially, the rotation with dividing can have no error angle, even when the A1 size documents is divided into 200 segments, whereas the conventional method cause the error angle of 1. 68 degrees.

In our current work, we attempt to make an automatic diagnostic system of lung cancer based on the analysis of the sputum color images. In order to form general diagnostic rules, we have collected a database with thousands of sputum color images from normal and abnormal subjects. As a first step, in this paper, we present a segmentation method of sputum color images prepared by the Papanicalaou standard staining method. The segmentation is performed based on an energy function minimization using an unsupervised Hopfield neural network (HNN). This HNN have been used for the segmentation of magnetic resonance images (MRI). The results have been acceptable, however the method have some limitations due to the stuck of the network in an early local minimum because the energy landscape in general has more than one local minimum due to the nonconvex nature of the energy surface. To overcome this problem, we have suggested in our previous work some contributions. Similarly to the MRI images, the color images can be considered as multidimensional data as each pixel is represented by its three components in the RGB image planes. To the input of HNN we have applied the RGB components of several sputum images. However, the extreme variations in the gray-levels of the images and the relative contrast among nuclei due to unavoidable staining variations among individual cells, the cytoplasm folds and the debris cells, make the segmentation less accurate and impossible its automatization as the number of regions is difficult to be estimated in advance. On the other hand, the most important objective in processing cell clusters is the detection and accurate segmentation of the nuclei, because most quantitative procedures are based on measurements of nuclear features. For this reason, based on our collected database of sputum color images, we found an algorithm for NonSputum cell masking. Once these masked images are determined, they are given, with some of the RGB components of the raw image, to the input of HNN to make a crisp segmentation by assigning each pixel to label such as Background, Cytoplasm, and Nucleus. The proposed technique has yielded correct segmentation of complex scene of sputum prepared by ordinary manual staining method in most of the tested images selected from our database containing thousands of sputum color images.

Recent progress in neural network research has demonstrated the usefulness of neural networks in a variety of areas. In this work, its application in the spatial resolution improvement of a remotely sensed low resolution thermal infrared image using high spatial resolution of visible and near-infrared images from Landsat TM sensor is described. The same work is done by an algebraic method. The tests developed are explained and examples of the results obtained in each test are shown and compared with each other. The error analysis is also carried out. Future improvements of these methods are evaluated.

In this paper, a two-dimensional (2-D) binary-valued (BV) lapped transform (LT) is proposed. The proposed LT has basis images which take only BV elements and satisfies the axial-symmetric (AS) property. In one dimension, there is no 2-point LT with the symmetric basis vectors, and the property is achieved only with the non-overlapping basis which the Hadamard transform (HT) has. Hence, in two dimension, there is no 22-point separable ASLT, and only 2-D HT can be the 22-point separable AS orthogonal transform. By taking non-separable BV basis images, this paper shows that a 22-point ASLT can be obtained. Since the proposed LT is similar to HT, it is referred to as the lapped Hadamard transform (LHT). LHT of larger size is shown to be provided with a tree structure. In addition, LHT is shown to be efficiently implemented by a lattice structure.

We introduce an image contour clustering method based on a multiscale image representation and its application to image compression. Multiscale gradient planes are obtained from the mean squared sum of 2D wavelet transform of an image. The decay on the multiscale gradient planes across scales depends on the Lipshitz exponent. Since the Lipshitz exponent indicates the spatial differentiability of an image, the multiscale gradient planes represent smoothness or sharpness around edges on image contours. We apply vector quatization to the multiscale gradient planes at contours, and cluster the contours in terms of represntative vectors in VQ. Since the multiscale gradient planes indicate the Lipshitz exponents, the image contours are clustered according to its gradients and Lipshitz exponents. Moreover, we present an image recovery algorithm to the multiscale gradient planes, and we achieve the skech-based image compression by the vector quantization on the multiscale gradient planes.

Image restoration using estimated parameters of image model and noise statistics is presented. The image is modeled as the output of a 2-D noncausal autoregressive (NCAR) model. The parameter estimation process is done by using the autocorrelation function and a biased term to a conventional least-squares (LS) method for the noncausal modeling. It is shown that the proposed method gives better results than the other parameter estimation methods which ignore the presence of the noise in the observation data. An appropriate image model selection process is also presented. A genetic algorithm (GA) for solving a multiobjective function with single constraint is discussed.

We have developed a new autofocus method using image processing techniques. This method consists of two steps. The first step is the preset of an objective lens condition with the aid of the feedback of Z-sensor. Next, a hole pattern to be measured is detected using the pattern recognition. In the second step, the E-beam is shifted to the center of a hole pattern and scanned across the axis of a pattern. The exciting current of the objective lens is changed at constant intervals, where the center position of the range is the preset value of the Z-sensor. The best focus condition is determined based on the signal profile obtained by the autofocus scan. The measurement repeatability (3σ) can be achieved within 3. 9 nm. The percentage of success of 98. 7% can be realized in the present autofocus method.

A function approximation scheme for image restoration is presented to resolve conflicting demands for smoothing within each object and differentiation between objects. Images are defined by probability distributions in the augmented functional space composed of image values and image planes. According to the fuzzy Hough transform, the probability distribution is assumed to take a robust form and its local maxima are extracted to yield restored images. This statistical scheme is implemented by a feedforward neural network composed of radial basis function neurons and a local winner-takes-all subnetwork.

Many image-processing techniques are based on texture features or gradation features of the image. However, Landsat images are complex; they also include physical features of reflection radiation and heat radiation from land cover. In this paper, we describe a method of constructing a super-resolution image of Band 6 of the Landsat TM sensor, oriented to analysis of an agricultural area, by combining information (texture features, gradation features, physical features) from other bands. In this method, a knowledge-based hierarchical classifier is first used to identify land cover in each pixel and then the least-squares approach is applied to estimate the mean temperature of each type of land cover. By reassigning the mean temperature to each pixel, a finer spatial resolution is obtained in Band 6. Computational results show the efficiency of this method.

In this letter, a design method of linear-phase paraunitary filter banks is proposed for an odd number of channels. In the proposed method, a non-linear unconstrained optimization process is assumed to be applied to a lattice structure which makes the starting guess of design parameters simple. In order to avoid insignificant local minimum solutions, a recursive initialization procedure is proposed. The significance of our proposed method is verified by some design examples.

We have developed and fabricated an LSI called the FMPP-VQ64. The LSI is a memory-based shared-bus SIMD parallel processor containing 64 PEs, intended for low bit-rate image compression using vector quantization. It accelerates the nearest neighbor search (NNS) during vector quantization. The computation time does not depend on the number of code vectors. The FMPP-VQ64 performs 53,000 NNSs per second, while its power dissipation is 20 mW. It can be applied to the mobile telecommunication system.

We introduce a concept of regularization into Genetic Algorithms (GAs). Conventional GAs include no explicit regularizing operations. However, the regularization is very effective in solving ill-posed problems. So, we propose a method of regularization to apply GAs to ill-posed problems. This regularization is a kind of consensus operation among neighboring individuals in GAs, and plays the role of `smoothing the solution. ' Our method is based on the evaluation of macroscopic fitness, which is a new fitness criterion. Conventional fitness of an individual in GAs is defined only from the phenotype of the individual, whereas the macroscopic fitness of an individual is evaluated from the phenotypes of the individual and its neighbors. We tested our regularizing operation by means of experiments with an elastic image mapping problem, and showed the effectiveness of the regularization.

This paper describes an automatic region segmentation method which is detectable nuclei regions from hematoxylin and eosin (HE)-stained breast tumor images using artificial organisms. In this model, the stained images are treated as virtual environments which consist of nuclei, interstitial tissue and background regions. The movement characteristics of each organism are controlled by the gene and the adaptive behavior of each organism is evaluated by calculating the similarities of the texture features before and after the movement. In the nuclei regions, the artificial organisms can survive, obtain energy and produce offspring. Organisms in other regions lose energy by the movement and die during searching. As a result, nuclei regions are detected by the collective behavior of artificial organisms. The method developed was applied to 9 cases of breast tumor images and detection of nuclei regions by the artificial organisms was successful in all cases. The proposed method has the following advantages: (1) the criteria of each organism's texture feature values (supervised values) for the evaluation of nuclei regions are decided automatically at the learning stage in every input image; (2) the proposed algorithm requires only the similarity ratio as the threshold value when each organism evaluates the environment; (3) this model can successfully detect the nuclei regions without affecting the variance of color tones in stained images which depends on the tissue condition and the degree of malignancy in each breast tumor case.