Avaliação da rugosidade superficial do solo utilizando técnicas de sensoriamento remoto e análise de imagens / Evaluation of soil surface roughness using remote sensing techniques and analysis of image

AUTOR(ES)
DATA DE PUBLICAÇÃO

2009

RESUMO

To study the top-soil rugosity using the remote sense techniques, we used one area of 1.200 m, divided on three blocks with five treatments each one, where the dimensions of the parcels were from 20 m of length by 3 m of wideness and 1 m between each parcel, and each block had an area of 20 x 18 m. In every block it was made an assortment of the localization of the parcels of the five treatments. The treatments consisted in five distinct ways to prepare the soil, where the plowing was made in all treatments. The harrow was used only in four treatments and in three treatments we used the rotavator with different regulations, providing different rugosities. To obtain the aerial images, we put two digital cameras in a balloon. One camera had colored images and the other was prepared to collect the infrared images. It was developed two circuits to shoot the cameras remotely. One circuit, named of base earth, where it was configured the way to shoot: manual or automatic. The other circuit was named remote base, where the cameras were plugged. We evaluated for heights: 4, 20, 50 e 100 m. The images obtained were processed with the co- occurrence matrix techniques where there were extracted eight texture descriptors of the images. We still evaluated the influence of the size of the blocks, removed from the image to classify the rugosity classes in the images. It was made a soil sampling to measure the moisture, texture and chemical analysis as a characterization of the soil. The treatment only with the plowing presented a higher rugosity index and the smaller was for the treatment with the plow, harrow and rotavator with the cover closed. The perfilometer doesnt distinguish the five rugosity classes, statistically. In relation to the size of the blocks of the image, the block with the bigger dimension, 250x250 pixels presented the higher values of the kappa index for the heights of 4 and 20 m. For the height of 50 m, the block 90 x 90 obtained the best result. The systems developed to obtain the images are completely practicable to use with the remote sense techniques, with the advantage to be cheaper. The angle orientation of the pixel to produce the co-occurrence matrix with a better performance in the classification was for 45 e 135. Only the bands B, R and IV B, for the height of 20 m, had the kappa index values the same as 1,0. The bands B, IV G and IV B had the tendency to present a higher % of the kappa index up than 0,90. The combination of the texture descriptors had the tendency to have higher values of the kappa index with the combination of 2, 3, 4 and 5 descriptors. The proposed classifier was considered reliable to study the top-soil rugosity. The classified distinguished the five classes of the top-soil rugosity.

ASSUNTO(S)

image classification digital camera câmera digital roughness classificação de imagens rugosidade mecanizacao agricola

ACESSO AO ARTIGO

Documentos Relacionados