Large Data Reduction Research Articles

Unique Shape Descriptor Algorithm for Medicinal Plant Identification (SDAMPI) With Abridged Image Database

In image processing, leaf shape recognition requires a huge database of similar images. Normal leaf image database construction requires more time and space. On the other hand, mobile offline applications are not able to contain huge image database with high pixel ratio. To solve this problem, mainly in medicinal plant identification, small leaf descriptors are necessary to completely provide the required plant information. Presently, the Botanists use shape reference table to recognize the following shapes of the leaf: ovate, cordate, elliptical, oblong, lanceolate and linear. As the shape numbers are invariant under scale, rotation and translation, which is highly desirable property for object recognition and chain code techniques preserve data by allowing large data reduction. Hence, in the proposed Shape Descriptor Algorithm for Medicinal Plant Identification (SDAMPI), we developed a descriptor to resolve the pixel selection issue of Freeman chain code and generate a unique leaf shape number. This leaf shape digital descriptor will act as a reference table for medicinal plant leaf shape identification. The performance of the proposed descriptor is evaluated through Jaccard similarity index graph and Levenshtein distance (LD) graph. From the results, it is confirming that, SDAMPI descriptor can detect Medicinal plant leaf shapes more accurately than existing methods.

A method for reduction of Acoustic Emission (AE) data with application in machine failure detection and diagnosis

Abstract The use of AE in machine failure diagnosis has increased over the last years. Most AE-based failure diagnosis strategies use digital signal processing and thus require the sampling of AE signals. High sampling rates are required for this purpose (e.g. 2 MHz or higher), leading to streams of large amounts of data. This situation is aggravated if fine resolution and/or multiple sensors are required. These facts combine to produce bulky data, typically in the range of GBytes, for which sufficient storage space and efficient signal processing algorithms are required. This situation probably explains why, in practice, AE-based methods consist mostly in the calculation of scalar quantities such as RMS and Kurtosis, and the analysis of their evolution in time. While the scalar-based approach offers the advantage of maximum data reduction; it has the disadvantage that most part of the information contained in the raw AE signal is lost unrecoverably. This work presents a method offering large data reduction, while keeping the most important information conveyed by the raw AE signal, useful for failure detection and diagnosis. The proposed method consist in the construction of a synthetic, unevenly sampled signal which envelopes the AE bursts present on the raw AE signal in a triangular shape. The constructed signal – which we call TriSignal – also permits the estimation of most scalar quantities typically used for failure detection. But more importantly, it contains the information of the time of occurrence of the bursts, which is key for failure diagnosis. Lomb-Scargle normalized periodogram is used to construct the TriSignal spectrum, which reveals the frequency content of the TriSignal and provides the same information as the classic AE envelope. The paper includes application examples in planetary gearbox and low-speed rolling element bearing.

Integrating high-content screening and ligand-target prediction to identify mechanism of action

High-content screening is transforming drug discovery by enabling simultaneous measurement of multiple features of cellular phenotype that are relevant to therapeutic and toxic activities of compounds. High-content screening studies typically generate immense datasets of image-based phenotypic information, and how best to mine relevant phenotypic data is an unsolved challenge. Here, we introduce factor analysis as a data-driven tool for defining cell phenotypes and profiling compound activities. This method allows a large data reduction while retaining relevant information, and the data-derived factors used to quantify phenotype have discernable biological meaning. We used factor analysis of cells stained with fluorescent markers of cell cycle state to profile a compound library and cluster the hits into seven phenotypic categories. We then compared phenotypic profiles, chemical similarity and predicted protein binding activities of active compounds. By integrating these different descriptors of measured and potential biological activity, we can effectively draw mechanism-of-action inferences.

Proper orthogonal decomposition of direct numerical simulation data: Data reduction and observer construction

In this paper, direct numerical simulation (DNS) data of an opposed-jet hydrogen/air diffusion flame are, in a postprocessing step, analyzed using the proper orthogonal decomposition (POD) technique. The aim of this work is twofold. The first goal is to compute a small number of space-dependent empirical eigenfunctions, so that a low-dimensional representation of the data generated by the large model of the discretized partial differential equations can be obtained using a weighted sum of these few eigenfunctions (POD modes). It is found that only six modes are needed for an accurate representation of the data in an extended range of inflow velocities. This large data reduction takes into account not only chemical kinetics but also transport phenomena in a full two-dimensional context and constitutes the first step toward the construction of low-dimensional dynamic models for the opposed-jet system. It is also found that the PODs have very good interpolatory properties. The second goal is to use part of the available data (i.e., partial measurements), together with the computed modes, to estimate, or, in the terminology of process control, to observe, the “unmeasured” quantities. It is found that only a small number of measurements are needed to obtain accurate estimates of the rest of the data.

Deterministic and Stochastic Separation of Digital Images

The digital image of a scene is composed of patterned and textured regions, which are separated by boundaries or edges. This paper presents a three stage approach to separating the regions which constitute the structural or deterministic component of the image and the textures which constitute the stochastic part of the image. The data dependent system methodology is extended to the boundary value problem, to achieve time (space) and frequency localization on the image data. The peculiarity of this approach is that it identifies each texture as one entity, as opposed to conventional approaches which give rise to a number of edgelets for each texture in the image. The first stage generates a mathematical characterization of the image using the intrinsic spatial dependencies of the image intensities. The dynamics of the image intensities is characterized by the Green's function of the adequate model. The components of intensities unexplained by the Green's function are captured in the model residuals. These residuals contain both boundary values and noise. Boundary values occur at the outlier residual locations and are nonzero only at the region boundaries. In the second stage a linear regression that minimizes the stochastic part of the model is used to estimate the boundary values, which are measures of edge strength. The third stage involves the deterministic and stochastic reconstruction of the image, which demonstrates the structural and textural separation of the image. Convolution of boundary values with Green's function gives the structural component, whereas convolution of white noise with Green's function produces the textural component of the image. The results show excellent structural textural separation in addition to large data reduction capability.

Automated 3-D cell population analysis in thick tissue sections using laser-scanning confocal-microscopy data

This paper presents methods for performing detailed quantitative automated three dimensional (3-D) analysis of cell populations in thick tissue sections while preserving the relative 3-D locations of cells. Specifically, the method disambiguates overlapping clusters of cells, and accurately measures the volume, 3-D location, and shape parameters for each cell. Finally, the entire population of cells is analyzed to detect patterns and groupings with respect to various combinations of cell properties. All of the above is accomplished with zero subjective bias.In this method, a laser-scanning confocal light microscope (LSCM) is used to collect optical sections through the entire thickness (100 - 500μm) of fluorescently-labelled tissue slices. The acquired stack of optical slices is first subjected to axial deblurring using the expectation maximization (EM) algorithm. The resulting isotropic 3-D image is segmented using a spatially-adaptive Poisson based image segmentation algorithm with region-dependent smoothing parameters. Extracting the voxels that were labelled as "foreground" into an active voxel data structure results in a large data reduction.

Holographic Data Reduction

Very large bandwidths are required for the transmission of holographic data for systems such as TV. This paper presents a technique in which the large bandwidths normally required are traded off for either increased noise or decreased resolution in the image. The light radiated from the object is diffracted by an intermediate dispersion medium and collected at the hologram aperture. Correct illumination of this hologram provides an image beam that passes back through the intermediate medium and comes to focus in the space originally occupied by the object. By proper selection of the dispersion medium, the hologram aperture can be made extremely small, thus representing a large data reduction. The three dimensionality of the image and the original viewing angles are maintained. Included in the paper are experimental results that show reconstructed images after a data reduction of as much as 3600.