Abstract The techniques of clustering and space transformation have been successfully used in the past to solve a number of pattern recognition problems. In this article, the authors propose a new approach to content‐based image retrieval (CBIR) that uses (a) a newly proposed similarity‐preserving space transformation method to transform the original low‐level image space into a high‐level vector space that enables efficient query processing, and (b) a clustering scheme that further improves the efficiency of our retrieval system. This combination is unique and the resulting system provides synergistic advantages of using both clustering and space transformation. The proposed space transformation method is shown to preserve the order of the distances in the transformed feature space. This strategy makes this approach to retrieval generic as it can be applied to object types, other than images, and feature spaces more general than metric spaces. The CBIR approach uses the inexpensive “estimated” distance in the transformed space, as opposed to the computationally inefficient “real” distance in the original space, to retrieve the desired results for a given query image. The authors also provide a theoretical analysis of the complexity of their CBIR approach when used for color‐based retrieval, which shows that it is computationally more efficient than other comparable approaches. An extensive set of experiments to test the efficiency and effectiveness of the proposed approach has been performed. The results show that the approach offers superior response time (improvement of 1–2 orders of magnitude compared to retrieval approaches that either use pruning techniques like indexing, clustering, etc., or space transformation, but not both) with sufficiently high retrieval accuracy.

A promising approach to content-based image retrieval, proposed by Choubey and Raghavan (1997), involves the representation of the original image space, in terms of low-level image features into a feature space, where images are represented as vectors of high-level features. The retrieval system based on that approach consists of three phases: database population; online addition; and image retrieval. Though their framework supports content-based retrieval of images, the issues that arise when database grows dynamically and user queries are different from those in database, were not investigated. In the current work, we: (i) experimentally investigate issues relating to online addition of new images and image retrieval; and (ii) provide a theoretical analysis of the complexity and effectiveness of our retrieval system by comparing it with the conventional approach. We have performed an extensive set of experiments to test the efficiency, effectiveness and scalability of our approach. The experimental results show that our approach is not only efficient but also effective in retrieving images even when the image database is dynamic and user queries are framed with images that are external to the database.

When signals from a user (fixed or mobile) are received at multiple antennas, the signals can be arrayed to increase the signal-to-noise ratio (SNR) of the received signal. This article describes two maximal ratio combining techniques referred to as full spectrum combining (FSC) and complex symbol combining (CSC), and then compares them in terms of the achievable arraying gain. Although the performance of these techniques is derived for the case of binary phase shift keying (BPSK) signalling in an additive white Gaussian noise channel, the analytical techniques here are applicable to other modulations and channels. The paper also addresses the use of symbol SNR degradation and symbol SNR loss as system performance measures. It is shown that degradation and loss are equal at weak signal levels but degradation is a lower bound for loss when the signal is strong. The telemetry arraying techniques described are applicable to any phase modulated signal. We assess the deep space communication channel performance scenario where the weakest signals ever are detected.

All-digital phase locked loops (DPLLs) have many advantages over analog loops. However, due to digital device limitations and costs, superwide PLLs with front-end bandwidths as high as one gigahertz are commonly implemented using analog parts. This article presents a new architecture that allows an all-digital implementation of superwide PLLs. The problem of operating digital components at high speed is avoided here (without reducing the front-end bandwidths) by inserting a multirate digital filter bank in front of the DPLL. The new design is shown to have steady-state and transient performance that is identical to a conventional DPLL.

This paper discussed four open-loop frequency acquisition techniques for suppressed-carrier BPSK signals. The techniques, which do not require accurate symbol timing nor a data preamble sequence, employ fast Fourier transforms (FFT's) on the Costas loop phase detector output to detect the Doppler frequency offset. Performance of the full-, half-, and staggered-symbol integration techniques as well as the low-pass filter technique are compared in terms of the error signal output SNR's. Then, the probability of detecting the frequency offset is computed for the special case when the frequency of the tone to be detected coincides with one of the FFT-bin center frequencies. It is shown that the performance of the integration techniques depends strongly on the symbol timing offset. On the other hand, the detection probability for the one-pole arm filter technique depends directly on the filter bandwidth. The staggered integration technique is shown to have comparable performance to the low-pass filter technique for low-to-medium symbol SNR's but the latter is superior at high symbol SNR's.>

The frequency and phase acquisition performance of three quadrature phase shift keying (QPSK) carrier tracking loops, the MAP estimation loop, the Costas crossover loop, and the generalized Costas loop, is described. Acquisition time and probability of acquisition as a function of both loop signal-to-noise ratio and frequency offset to loop bandwidth ratio are obtained via computer simulations for type II and III loops. It is shown that the MAP loop, which results in the smallest squaring loss for all signal-to-noise ratios, is sometimes outperformed by the other two loops in terms of acquisition time and acquisition probability.>