Non-destructive sensing for determining Sunagoke moss water content -bio-inspired approaches-

Abstract

One of the primary determinants of Sunagoke moss Rachomitrium japonicum growth is water availability. There is need to develop non-destructive sensing of Sunagoke moss water content to realize automation and precision irrigation in a close bio-production system. Machine vision can be utilized as non-destructive sensing to recognize changes in some kind of features that describe the water conditions from the appearance of wilting Sunagoke moss. The goal of this study is to propose and investigate bio-inspired algorithms i.e. neural-genetic algorithms (neural-GAs) and neural-ant colony optimization (neural-ACO) to find the most significant set of image features suitable for predicting cultured Sunagoke moss water content in a close bio-production system. Features extracted consisted of 13 colour features, 90 textural features (grey level co-occurrence matrix, RGB, HSV and HSL colour co-occurrence matrix textural features) and three morphological features. Each colour space consisted of ten textural features algorithms: entropy, energy, contrast, homogeneity, sum mean, variance, correlation, maximum probability, inverse difference moment and cluster tendency. The specificity of this problem was that we were not looking for single image feature but several associations of image features that may be involved in determining water content of Sunagoke moss. Neural-ACO had better prediction performance with lower number of features than neural-GAs. The minimum validation prediction mean square error (MSE) achieved was 2.02x10^-3 when using 10 relevant features.