Genetic algorithm optimization of adaptive multi-scale GLCM features

We introduce a new second-order method of texture analysis called Adaptive Multi-Scale Grey Level Co-occurrence Matrix (AMSGLCM), based on the well-known Grey Level Co-occurrence Matrix (GLCM) method. The method deviates significantly from GLCM in that features are extracted, not via a fixed 2D weighting function of co-occurrence matrix elements, but by a variable summation of matrix elements in 3D localized neighborhoods. We subsequently present a new methodology for extracting optimized, highly discriminant features from these localized areas using adaptive Gaussian weighting functions. Genetic Algorithm (GA) optimization is used to produce a set of features whose classification worth is evaluated by discriminatory power and feature correlation considerations. We critically appraised the performance of our method and GLCM in pairwise classification of images from visually similar texture classes, captured from Markov Random Field (MRF) synthesized, natural, and biological origins. In these cross-validated classification trials, our method demonstrated significant benefits over GLCM, including increased feature discriminatory power, automatic feature adaptability, and significantly improved classification performance.

The influence of coating materials on some properties of alginate beads and survivability of microencapsulated probiotic bacteria

The probiotics, Lactobacillus acidophilus 547, Bifidobacterium bifidum ATCC 1994, and Lactobacillus casei 01, were encapsulated into uncoated calcium alginate beads and the same beads were coated with three types of material, chitosan, sodium alginate, and poly-L-lysine in combination with alginate. The thickness of the alginate beads increased with the addition of coating materials. No differences were detectable in the bead strength by texture analysis or in the thickness of the beads with different types of coating materials by transmission electron microscopy. The survivability of three probiotics in uncoated beads, coated beads, and as free cells (unencapsulated) was conducted in 0.6% bile salt solution and simulated gastric juice (pH 1.55) followed by incubation in simulated intestinal juice with and without 0.6% bile salt. Chitosan-coated alginate beads provided the best protection for L. acidophilus and L. casei in all treatments. However, B. bifidum did not survive the acidic conditions of gastric juice even when encapsulated in coated heads. (C) 2004 Elsevier Ltd. All rights reserved.

Hydraulic properties of layered soils influence survival of Rhodes grass (Chloris gayana Kunth.) during water stress

Survival of vegetation on soil-capped mining wastes is often impaired during dry seasons due to the limited amount of water stored in the shallow soil capping. Growth and survival of Rhodes grass (Chloris gayana) during soil drying on various layered capping sequences constructed of combinations of topsoil, subsoil, seawater-neutralised residue sand and low grade bauxite was determined in a glasshouse. The aim was to describe the survival of Rhodes grass in terms of plant and soil water relationships. The soil water characteristic curve and soil texture analysis was a good predictor of plant survival. The combination of soil with a high water holding capacity and low soil water diffusivity (e.g. subsoil with high clay contents) with soil having a high water holding capacity and high diffusivity (e.g. residue sand) gave best survival during drying down (up to 88 days without water), whereas topsoil and low grade bauxite were unsuitable (plants died within 18-39 days). Clayey soil improved plant survival by triggering a water stress response during peak evaporative water demand once residue sand dried down and its diffusivity fell below a critical range. Thus, for revegetation in seasonally dry climates, soil capping should combine one soil with low diffusivity and one or more soils with high total water holding capacity and high diffusivity.

Simple texture-based liberation modelling of ores

Mineralogical analysis is often used to assess the liberation properties of particles. A direct method of estimating liberation is to actually break particles and then directly obtain liberation information from applying mineralogical analysis to each size-class of the product. Another technique is to artificially apply random breakage to the feed particle sections to estimate the resultant distribution of product particle sections. This technique provides a useful alternative estimation method. Because this technique is applied to particle sections, the actual liberation properties for particles can only be estimated by applying stereological correction. A recent stereological technique has been developed that allows the discrepancy between the linear intercept composition distribution and the particle section composition distribution to be used as guide for estimating the particle composition distribution. The paper will show results validating this new technique using numerical simulation. (C) 2004 Elsevier Ltd. All rights reserved.