ABSTRACT
	The land management and analysis requires the use of information acquired as digital images by aerial or spatial 
sensors. The characterization of landscape elements can be focused studying either the spectral or spatial information, 
that is, their shapes, sizes, grey levels distribution, etc. The objective of texture analysis is to study the spatial 
information and distribution of grey levels in order to classify the images. In the present work, an analysis of textures 
is carried out for classifying different types of landscapes using aerial and satellite images containing different 
coverages to determine the most suitable variables. These images depict natural forest covers in mountain environments 
(Sierra de Espadán), mixed covers of natural vegetation and crops (Menorca and Valle de Ayora) and horticulture crops, 
anthropic and peri-urban areas of Valencia.
	The analysis of textures can be focused from several points of view: statistical, structural, based on models, 
transform-based, etc. Given the good results obtained using some statistical methods and the good perspectives that present 
the Wavelet Transform, we applied these two methods in order to determine the textural patterns of a set of images. The 
analysed variables are the first and second order statistics of the Grey Level Co-occurrences Matrix (GLCM), and others 
obtained from the histogram.
	The Wavelet Transform presents a good behaviour in the study of the space-frequency relation. It is based on 
mathematical functions that decompose data or signals in different frequency components, that are studied with a resolution 
adapted to their particular scale. The wavelet functions allow the design of filter banks that are applied first along the 
rows with a dyadic downsampling, and then along the columns with a second downsampling with the same dyadic factor, obtaining
the information of high and low frequencies of the image. Several parameters are analysed in this transform, such as the 
families and supports, algorithms of application, levels and scales of decomposition, etc.
	The results obtained in the present work show the importance of a suitable selection of the textural variables and 
the specific parameters of the Wavelet Transform to be used. There is also an important improvement in the accuracy of the 
classification using the variables obtained from this transform. In urban, agricultural and forest environments is very 
important to differentiate between kinds of coverages and densities of occupation. For this objective, the textural 
variables obtained from the wavelet transform complete the statistical variables of the panchromatic image. This method 
is more suitable when the classes are homogeneous and not very separable, yielding better results. On the other hand, the 
border effect is reduced by means of some algorithms for extraction and analysis of the external areas of the different 
textures.