• Tidak ada hasil yang ditemukan

Using Texture Analysis of Airborne Imagery and L-Systems Modeling for Plant Cover Classification. (5254)

N/A
N/A
Protected

Academic year: 2024

Membagikan "Using Texture Analysis of Airborne Imagery and L-Systems Modeling for Plant Cover Classification. (5254)"

Copied!
1
0
0

Teks penuh

(1)

Using Texture Analysis of Airborne Imagery and L-Systems Modeling for Plant Cover Classification. (5254)

Authors:

L. Pachepsky* - USDA-ARS, Beltsville, MD C. Walthall - USDA-ARS, Belstville, MD C. Daughtry - USDA-ARS, Beltsville, MD M. Kaul - USDA-ARS, Beltsville, MD

Abstract:

In computer graphics, computer vision, and pattern recognition, the texture is a leading feature of any image analysis and synthesis. Many publications in these fields show that the texture analysis can be an extremely useful approach in the remote sensing imagery analysis. However it is rarely used in analysis of the remote sensing data in agriculture, even though the basic GIS software (e.g., IDRISI) has all the tools necessary for texture analysis except some most complex ones, like wavelets and Gabor’s

functions. The latter algorithms can be easily implemented using such wildly available packages as MatLab. Texture is a natural and inherent characteristic of any plant cover, and different crops or natural plant covers have different quantitative texture characteristics. Texture characteristics of about 40 images of various agricultural and natural plant covers containing a single plant species and

photographed from various distances as well as photographs of L-models of various plants have been analyzed as a training set. Then the algorithm developed on this set was used for classification and segmentation of airborne images containing a variety of species. Classification was successful in about 80% of cases and segmentation was acceptably successful in about 70% of calculations. The ways to improve these results are discussed.

Speaker Information: Ludmila Pachepsky, USDA-ARS, Beltsville, MD, HRSL-ARS-USDA, Bldg 007 Rm 125, 10300 Baltimore Ave, BARC-West, Beltsville, MD 20708; Phone: 301 504 5138; E-mail:

[email protected]

Session Information: Monday, November 1, 2004, 4:00 PM-6:00 PM Presentation Start: 4:00 PM (Poster Board Number: 0221) Keywords: Classification; Imagery; Texture; L-systems

Referensi

Dokumen terkait