Analiza umidităţii argilei proaspete prin tehnici de procesare a imaginii
Moisture analysis of raw clay using image processing

Author(s): Alexandrina-Elena Andon
Subject(s): Architecture
Published by: INCD URBAN-INCERC
Keywords: raw clay; soil moisture; artificial neural networks; image-processing
Summary/Abstract: Clay can absorb and release moisture faster and to a greater extent than any other building material. The normal moisture content that unsaturated clay can absorb is 5% to 7% by weight. Numerous studies have focused on determining soil moisture in a practical, fast, non-destructive and relatively inexpensive way by evaluating the effectiveness of artificial neural networks (ANNs). The image processing method for determining moisture involves using artificial neural networks to accurately determine the moisture content of soils, based on the assumption that soils change color with water content. Disturbed samples of raw clay taken from the Vaslui region, Codăești village, were used for training and testing ANN. Input data were collected in the laboratory from samples subjected to different water contents, consisting of color photographs, taken with a 24.1 megapixel digital camera. The higher the water content, the darker the sample's color. A multi-layer neural network was created with 3 neurons in the input layer, each neuron representing the pixel area for the three colors (red, green and blue) resulting from the segmentation of the image into regions, with a single hidden layer and a single output layer. The ANN output consisted of the soil moisture value, determined for each soil sample by classical methods.

• Details
• Contents