Techniques of Digital Image Processing to indicate and quantify the area decrease in reservoir in São Paulo because of the dry process
ISBN 978-85-88783-11-9
Authors
2Carolina, C.; 2Chaves, C.; 2Erivaldo, S.; 2Santos, A.
2UNESP PRESIDENTE PRUDENTE Email: carolina.dias.chaves@gmail.com
2UNESP Email: carolina.dias.chaves@gmail.com
2UNESP Email: silva.erivaldo@gmail.com
2UNESP Email: amandafructuozodossantos@gmail.com
Abstract
In the area of cartography, studies correlated with Mathematical Morphology (MM) and Digital Image Processing (DIP) assist in the application of Remote Sensing. These in turn seek to heal the deficiency of cartographic products. From the studies conducted on research and using the techniques of the respective areas, this paper aims to detect the contours of the Cantareira dam (located in the state of São Paulo) in two images. An image corresponds to the area of study in 2013 with the tank showing a normal water level and in 2014, with the reservoir in a critical state concerning the volume of water. The difference between the results obtained for the contour of the reservoir in these two images corresponds to a drastic reduction in the volume of water resulting from the long period of drought in the state of São Paulo, resulting a severe supply problem. In order to compare the outline of the dam, two softwares have been used, Matlab along with the Mathematical Morphology Toolbox, which was used for detecting the contour of the reservoir through the use of routine, and the CARTOMORFH software (software developed in FCT - UNESP), in which the function was implemented to detect the difference in area between the contours of the reservoir at normal and dry periods alike. To detect the contour of the reservoir we used mmreadgray operators, responsible for opening the original image when grayscale was applied; the mmhdome, which smoothes the image to facilitate further processing; the mmneg, used for inversion of the gray levels of the pixels; the mmbinary, responsible for turning an image into grayscale and binary; and the mmareaopen, aiming to remove the image segmentation, all tones below the threshold value were removed with this last operator.The set of operators resulted in a routine capable of detecting the contour of the Cantareira reservoir. After the shell contour detection, the CARTOMORPH software was used to compare the resulting images in both study periods to quantify the areas and the difference between them. A cmDif routine was implemented, which has the function of generating an image difference between the two results of the contour of the reservoir used in the two images. The reported results prove the fall of the water level in the reservoir. We concluded that the drop in water level in the reservoir in 2014, can be proven by the decrease of the reservoir area that was approximately 783600m2. The value was calculated from the result of the difference in pictures, in which each pixel was considered an area of 30m2. Thus the aim of the work was to identify approximately a decrease in the boundary area of the reservoir, and this decrease indicates a drastic reduction in the volume of water in the reservoir. This result may support the planning area to minimize or indicate solutions to the water supply of cities to be taken up at appropriate levels.