Abstract: Interest in multiresolution techniques for signal processing and analysis is increasing steadily. An important instance of such a technique is the so-called pyramid decomposition scheme. This paper presents a general theory for constructing linear as well as nonlinear pyramid decomposition schemes for signal analysis and synthesis. The proposed theory is based on the following ingredients: 1) the pyramid consists of a (finite or infinite) number of levels such that the information content decreases toward higher levels and 2) each step toward a higher level is implemented by an (information-reducing) analysis operator, whereas each step toward a lower level is implemented by an (information-preserving) synthesis operator. One basic assumption is necessary: synthesis followed by analysis yields the identity operator, meaning that no information is lost by these two consecutive steps. Several examples of pyramid decomposition schemes are shown to be instances of the proposed theory: a particular class of linear pyramids, morphological skeleton decompositions, the morphological Haar pyramid, median pyramids, etc. Furthermore, the paper makes a distinction between single-scale and multiscale decomposition schemes, i.e., schemes without or with sample reduction. Finally, the proposed theory provides the foundation of a general approach to constructing nonlinear wavelet decomposition schemes and filter banks.

Abstract: The advent of large image databases (>10000) has created a need for tools which can search and organize images automatically by their content. This paper focuses on the use of hierarchical tree-structures to both speed-up search-by-query and organize databases for effective browsing. The first part of this paper develops a fast search algorithm based on best-first branch and bound search. This algorithm is designed so that speed and accuracy may be continuously traded-off through the selection of a parameter /spl lambda/. We find that the algorithm is most effective when used to perform an approximate search, where it can typically reduce computation by a factor of 20-40 for accuracies ranging from 80% to 90%. We then present a method for designing a hierarchical browsing environment which we call a similarity pyramid. The similarity pyramid groups similar images together while allowing users to view the database at varying levels of resolution. We show that the similarity pyramid is best constructed using agglomerative (bottom up) clustering methods, and present a fast sparse clustering method which dramatically reduces both memory and computation over conventional methods.

Abstract: Presented is a new 80 GOPS video-processing chip capable of performing video rate vision applications. These applications include real-time video stabilization, mosaicking, video fusion, motion-stereo and video enhancement. The new vision chip, code-named Acadia, is the result of over 15 years of research and development by Sarnoff in the areas of multi-resolution pyramid-based vision processing and efficient computational architectures. The Acadia chip represents the third generation of ASIC technology developed by Sarnoff, and incorporates the processing functions found in Sarnoff's earlier PYR-1 and PYR-2 pyramid processing chips as well as numerous other functions found in Sarnoff-developed video processing systems, including the PVT200. A demonstration board is being implemented and includes two video decoders, a video encoder and a PCI interface.

Abstract: We present a wavelet-based, high performance, hierarchical scheme for image matching which includes (1) dynamic detection of interesting points as feature points at different levels of subband images via the wavelet transform, (2) adaptive thresholding selection based on compactness measures of fuzzy sets in image feature space, and (3) a guided searching strategy for the best matching from coarse level to fine level. In contrast to the traditional parallel approaches which rely on specialized parallel machines, we explored the potential of distributed systems for parallelism. The proposed image matching algorithms were implemented on a network of workstation clusters using parallel virtual machine (PVM). The results show that our wavelet-based hierarchical image matching scheme is efficient and effective for object recognition.

Abstract: We describe a multiscale pyramid of line segments and develop algorithms which exploit that pyramid to recover image features - lines, curves, and blobs - from very noisy data.

Abstract: We propose the use of spectral analysis on certain noncommutative finite groups in digital signal processing and, in particular, image processing. We pay significant attention to groups constructed as wreath products of cyclic groups. Within this large class of groups, our approach recovers the discrete Fourier transform (DFT), the Haar wavelet transform, various multichannel pyramid filter banks, and other aspects of multiresolution analysis as special cases of a more general phenomenon. In addition, the group structure provides a rich algebraic structure that can be exploited for the analysis and manipulation of signals. Our approach relies on a synthesis of ideas found in the early work of Holmes (1987, 1990), Karpovsky and Trachtenberg (1985), and others on noncommutative filtering, as well as Diaconis's (1989) spectral analysis approach to understanding data.

Abstract: The two dimensional sample rate conversion capabilities of a video display system are used to produce three-dimensional effects. Linear and non-linear scaling is applied to a video image to convey a sense of depth. The three dimensional effects are used to increase the visual appeal of existing and new feature sets in display systems. A multi-faceted object representation, such as a representation of a cube or a pyramid, can be used to display different video images on each facet of the representation. By appropriately scaling each image on each facet, an impression of depth is achieved. The images on the different facets can be selected to represent different aspects of a common theme, such as datacast information related to a primary source of information. Channel changing on a television can be presented as a rotation of the multifaceted object. In like manner, other familiar representations, such as a representation of a book can be used, wherein channel changing is presented as a turning of the pages of the book, each television program being presented on a different page. Advanced features, such as program categorization, can be represented as tabs on the book that facilitate the selection of a particular category. In like manner, a rotation of a multifaceted object about one axis may correspond to a change of channel within a select category, whereas a rotation about another axis may correspond to a change of category. Techniques are presented for achieving these three dimensional effects with calculations that are well suited for execution via the sample rate converters of conventional display systems.

Abstract: A new technique for fast pattern recognition using normalized grey-scale correlation (NGC) is described. While NGC has traditionally been slow due to computational intensity issues, the introduction of both a pyramid structure and a local estimate of the correlation surface gradient allows for recognition in 10-50 ms using modest microcomputer hardware. The algorithm is designed to analyze the target off-line prior to starting the search. Issues surrounding determining an appropriate depth for the pyramid representation and performing sub-pixel localization of the target instance are discussed. The speed and robustness of the method makes it attractive for industrial applications.

Abstract: Digital watermarking is a promising technique to help protect data security and intellectual property rights. A blind digital image/video watermarking technique is proposed. A multiresolution pyramid transform is applied to a host image. A tight upper bound for the modulation of watermark signals is derived to control the fidelity loss. The robustness is optimized under the fidelity constraints. An optimum decision rule for the detection is derived using the statistical model of the generalized Gaussian distribution. Asymptotic performance analysis is provided. The simulations demonstrate that near perfect invisibility, very strong robustness, and optimum detection are obtained. The watermarking system can be applied to copyright notification, enforcement, and fingerprinting.

Abstract: Feature correspondence between images using an image pyramid is disclosed. In one embodiment, a fundamental matrix between a first and a second image is generated from the image pyramid. The image pyramid is first generated, and has a predetermined number of fineness levels, from a coarsest to a finest level. Each of the images has significant features at each pyramid level. A plurality of hypotheses, or particles, is generated for the fundamental matrix at the coarsest level, based on matching significant features of the images at the coarsest level. In an iterative procession through the levels of the image pyramid, starting at a present level initially set to the coarsest level and then subsequently advanced by one fineness level upon each iteration, an importance sampling function is first formulated from the hypotheses. The plurality of hypotheses is then generated at the next pyramid level based on the function, and on the significant features of the images at this next level. The iteration is complete when the next level has reached the finest pyramid level. The hypotheses generated at the finest level encapsulate the fundamental matrix together with its uncertainty.

Abstract: The demand for efficient communication and data storage is continuously increasing and signal representation and compression are important factors in digital communication and storage systems.This work deals with Frame based signal representation and compression. The emphasis is on the design of frames suited for efficient representation, or for low bit rate compression, of classes of signals.Traditional signal decompositions such as transforms, wavelets, and filter banks, generate expansions using an analysis-synthesis setting. In this thesis we concentrate on the synthesis or reconstruction part of the signal expansion, having a system with no explicit analysis stage. We want to investigate the use of an overcomplete set of vectors, a frame or an overcomplete dictionary, for signal representations and allow sparse representations. Effective signal representations are desirable in many applications, where signal compression is one example. Others can be signal analysis for different purposes, reconstruction of signals from a limited observation set, feature extraction in pattern recognition and so forth.The lack of an explicit analysis stage originates some questions on finding the optimal representation. Finding an optimal sparse representation from an overcomplete set of vectors is NP-complete, and suboptimal vector selection methods are more practical. We have used some existing methods like different variations of the Matching Pursuit (MP) [52] algorithm, and we developed a robust regularized FOCUSS to be able to use FOCUSS (FOCal Underdetermined System Solver [29]) under lossy conditions.In this work we develop techniques for frame design, the Method of Optimal Directions (MOD), and propose methods by which such frames can successfully be used in frame based signal representation and in compression schemes. A Multi Frame Compression (MFC) scheme is presented and experiments with several signal classes show that the MFC scheme works well at low bit rates using MOD designed frames. Reconstruction experiments provides complimentary evidence of the good properties of the MOD algorithm.

Abstract: We introduce a new method for automatic selection of optimal views in a shape-based method of multi-view 3D object representation and recognition. A 3D object is recognised by an optimum number of images taken from different views. The object boundary of each view is considered as a 2D shape and is represented by the locations of the maxima of its curvature scale space (CSS) image contours. An unknown object is then recognised by a single image taken from an arbitrary viewpoint. The method has been tested on a collection of 3D objects consisting of 15 aircraft of different shapes. Each object has been modelled using an optimised number of silhouette contours obtained from different view points. The results obtained show that robust and efficient 3D free form object recognition through multi-view representation can be achieved using the CSS representation even for large database retrieval applications.

Abstract: Text block segmentation is necessary in document layout analysis. An algorithm and its implementation that segregates text block by block (a block is either a title or a paragraph) from the provided document, e.g. newspaper image, based on pyramid structure is described in this paper. The pyramid structure, which is amenable for parallel processing on output, is a multi-resolution image representation. The pyramid structure also simulates what the human eyes see the document from afar visualizing the block structure of the document, the block segmentation can identify the titles, and distinguish different paragraphs based on the indentation between them. Our implementation will be used in a news articles retrieval project.

Abstract: A digital convolution apparatus, includes means (12) for determining a real-valued discrete-frequency representation of a desired digital filter and means (14) for transforming said discrete-frequency representation into a corresponding discrete-time representation. Means (16) circularly shift the discrete-time representation and means (18) apply a window to the discrete-time representation to produce a zero-padded reduced length filter. Thereafter means (24-36) convolve an input signal x[n] with the reduced length filter.

Abstract: A paintbrush-like image transformation is proposed. It is based on a random searching to insert brush-strokes into a generated image at decreasing scale of brush-sizes, without predefined models or interaction. One of the goals of the method is to transform the image into a representation that is very similar to the human sensation of artistic images. We introduce a sequential multiscale image decomposition method, based on simulated rectangular-shaped paintbrush strokes. The resulting images look like good-quality paintings with well-defined contours, at an acceptable distortion compared to the original image. The image can be described with the parameters of the consecutive paintbrush strokes, resulting in a parameter-series that can be used for compression. The painting process can be used for scale-space image representation, segmentation and contour detection, and image representation for retrieval purposes.

Abstract: PROBLEM TO BE SOLVED: To provide a correlation calculation method for calculating the distribution of distances up to an object at a high speed. SOLUTION: The image of an object is picked up by a pair of image pickup devices 10 and 11 and that image is made into a multi-resolution image in a pyramid structure by a multi-resolution image generating part 20. Next, the images of respective layers are made into edge images buy a spatial filter 30 for reducing an arithmetic quantity. Next, stereo collation is performed between the respective layers by a correlation operating part 40. At such a time, first of all, it is performed between the images, which are low-resolution images, of an upper layer and almost parallax is found. Besides, a correlation area and a search area corresponding to the stereo collation of the next layer are set by that parallax. By repeating such operation, the stereo collation to the final layer can be completed in a short time. Since the final layer is a source image, it becomes a high-density distance image equal with conventional one. Then, the distribution of distances to the object and the direction of a plane to form the object are calculated at a high speed by a distance distribution calculating part 50 and a plane direction estimating part 60.

Abstract: A method for image analysis, representation and re-synthesis is introduced. Unlike other schemes it is not pixel based but rather represents a picture as vector data, from which an altered version of the original image can be rendered. Representing an image as vector data allows performing operations such as zooming, retouching or colourising, avoiding common problems associated with pixel image manipulation. This paper brings together methods from the areas of computer vision, image compositing and image based rendering to prove that this type of image representation is a step towards accurate and efficient image manipulation.

Abstract: The image representation method has a large number of individual image elements visualized simultaneously in a grid consisting of columns and lines, each image element representing partial information of a scene/object from several views, with adjacent image elements providing light of different wavelengths or wavelength ranges. Also included are Independent claims for the following: (a) a three-dimensional image representation device; (b) a manufacturing method for a wavelength filter array

Abstract: . This paper describes an unsupervised algorithm for estimating the 3D profile of potholes in the highway surface, using structured illumination. Structured light is used to accelerate computation and to simplify the estimation of range. A low-resolution edge map is generated so that further processing may be focused on relevant regions of interest. Edge points in each region of interest are used to initialise open, active contour models, which are propagated and refined, via a pyramid, to a higher resolution. At each resolution, internal and external constraints are applied to a snake; the internal constraint is a smoothness function and the external one is a maximum-likelihood estimate of the grey-level response at the edge of each light stripe. Results of a provisional evaluation study indicate that this automated procedure provides estimates of pothole dimension suitable for use in a first, screening, assessment of highway condition.

Abstract: PROBLEM TO BE SOLVED: To provide an image compression device and method that are proper for a reduction in image memory capacity and a reduction in cost in image compression processing, and improve the speed of the processing. SOLUTION: The image compression processing device 1 subjects original image data to filter processing and compression processing depending upon a given image reduction rate before outputting image data. The original image data are transferred in units of a plurality of segment blocks, and a filter processing means 6 is provided for horizontal or vertical filter processing on the data. The filter processing divides the horizontal or vertical filter into a plurality of segment filters so that an object block and adjacently located blocks can be each independently subjected to the filter processing. Out of pixel data after the filter processing, pixel data in a pixel position specified according to the image reduction rate are selectively outputted.

Abstract: The algorithm developed uses an octree pyramid in which noise is reduced at the expense of the spatial resolution. At a certain level an unsupervised clustering without spatial connectivity constraints is applied. After the classification, isolated voxels and insignificant regions are removed by assigning them to their neighbours. The spatial resolution is then increased by the downprojection of the regions, level by level. At each level the uncertainty of the boundary voxels is minimised by a dynamic selection and classification of these, using an adaptive 3D filtering. The algorithm is tested using different data sets, including NMR data.

Abstract: In this paper, a kind of practical image segmentation algorithm for segmenting the license plate from its vehicle body background is presented, based on wavelet multiscale analysis and mathematical morphology. By multi resolution analysis and pyramid decomposition, the edge components with different spatial resolutions and different directions can be acquired, among which, the horizontal low frequency and vertical high frequency component has the most distinguished texture feature standing for the object region of the license plate. Then by further morphological operations, the noise in these edge components can be greatly decreased and the locations of the license plates can be also accurately determined. We have completed a series of experiments under different conditions. The experiment results show that the segmentation effects and it's orientation accuracy are improred significantly, so the proposed method is very effective for segmenting and locating noisy vehicle license plates.

Abstract: An adaptive multi-modal motion estimation algorithm for video compression builds a luminance pyramid for each image of a moving image sequence. From the top level image of the luminance pyramid a global motion vector is determined between images at times t and t+n. The global motion vector is used as a pivot point and to define a search area. For each block of a current top level image a search for a match is carried out around the pivot point within the search area. The resulting block motion vectors serve as initial conditions for the next higher resolution level. A refinement process results in a displaced frame difference value (DFD) for each block as an error measure. If the error measure is small, the motion vector is chosen as the motion vector for the current block. If the error measure is large, then a search within the search area around a zero motion pivot point is conducted. The motion vector that results in the smallest error measure is chosen as the motion vector for the current block. The refinement and zero pivot searches are repeated for each level down to the full resolution base of the pyramid, resulting in the desired estimated motion vectors for the image.

Abstract: To survey vegetation cover, obtaining reflectance temporal profiles is very important. Regularly and frequently acquired measurements are necessary to get "continuous" profiles. The SPOT4-VEGETATION and NOAA-AVHRR sensors fulfil these conditions. The major drawback is their coarse resolution of about 1 km. On the contrary, high spatial resolution sensors as SPOT 4-HRVIR (10 and 20 m), provide adequate images with a high probability of having only one kind of vegetation in one pixel. However, only a few images per year are available: this can't allow an accurate vegetation monitoring. Combining those two kinds of complementary sensors may provide efficient data for setting up temporal profiles. In this article a morphological pyramid approach is used to combine the high and coarse resolution data. In this paper, data fusion is based on the morphological pyramid principle in order to obtain the best filter.

Abstract: This paper discusses the usefulness of a partition-based image representation in the context of user-assisted segmentation, and compares it with the two other most common image representations in this framework: pixel-based and transition-based. Partition-based image representations allow user-interaction at the region level, which is a very natural and friendly manner to interact with images. The paper describes various region-based implementations of the most common types of user interaction; namely, initial object selection, object refinement, imposing object characteristics, selection of specific objects, selection of objects that are similar to a previously selected one. In all cases, partition-based image representations, and their associated region-based tools for user interaction, show up to be very suitable for interactive image segmentation.

Abstract: Computer vision and image processing techniques have been found increasingly useful for the fruit automatic quality inspection and defect sorting operation. However, real time fruit surface defect inspection and recognition is still a challenging project due to its complexity. In this paper, a fast approach for box dimension estimation based on a dual pyramid data structure is developed. Utilizing traditional fractal dimension and 4 oriented fractal dimensions as input values, a BP neural network is designed for identifying fruit defect area and stem, calyx concave area. The results of experiment show that the approach is effective for real time defect identification and is accurate. The rate of correct classification is 93% and the executing time of microcomputer for recognition of one undefined blob on the surface of apple is 4~7ms.

Abstract: A method and an apparatus for estimating the upper frequency band coefficients solely from low frequency information in a subband multiresolution decomposition In operation, a Bayesian classifier predicts the significance or insignificance of a high frequency signal, then a neural network estimates the sign and magnitude of the visually significant information and effectively eliminates insignificant information Finally an algorithm estimates upper frequencies based on lower frequencies within the same band This estimation is performed recursively for each level of a multiresolution decomposition pyramid until a reconstructed version of the data product is returned

Abstract: The measurement of perceptual similarities between textures is a difficult problem in applications such as image classification and image retrieval in large databases. Among the various texture analysis methods or models developed over the years, those based on a multi-scale multi-orientation paradigm seem to give more reliable results with respect to human visual judgement. This work introduces new texture features extracted from an oriented multi-scale pyramid structure called a "steerable pyramid". These texture features are then used in the search through an image database to find the most "similar" textures to a selected one. We have also introduced a relevance feedback to improve the retrieval quality.