Comparative Evaluation of Radial Basis Function Network Transfer Function for Filtering Speckle Noise in Synthetic Apert
Synthetic aperture radar (SAR) is imaging radar. It uses the movement of the antenna to synthesize a very large antenna and produces good resolution images of the mapped area. SAR images are polluted with speckle noise. Speckle noise inherently exists in
- PDF / 229,537 Bytes
- 10 Pages / 439.37 x 666.142 pts Page_size
- 95 Downloads / 192 Views
Abstract Synthetic aperture radar (SAR) is imaging radar. It uses the movement of the antenna to synthesize a very large antenna and produces good resolution images of the mapped area. SAR images are polluted with speckle noise. Speckle noise inherently exists in all types of coherent imagery including SAR. It reduces the quality of the image and causes serious difficulty in automatic interpretation of the images. The radial basis function network (RBFN) is gaining importance in speckle noise filtering. RBFN is an artificial neural network (ANN) that has only one hidden layer known as the radial center. In this paper we have used Gaussian function, thin plate spline function, quadratic function, and inverse quadratic function as radial centers to design and implement different RBFNs. The performance of these RBFNs for filtering speckle noise is evaluated based on mean square error(MSE) achieved during training and peak signal-to-noise ratio (PSNR).
Keywords Speckle Synthetic aperture radar filtering Artificial neural network
Radial basis
function
Noise
1 Introduction Synthetic aperture radar produces high-resolution two-dimensional images of mapped areas [1]; it is usually fitted on aircraft or spacecraft. A SAR works by illuminating the scanned surface with a beam of coherent electromagnetic radiation in a side-looking direction; the returned echoes from the illuminated surface are Khwairakpam Amitab (&) Debdatta Kandar A.K. Maji Department of Information Technology, North-Eastern Hill University, Shillong 793022, Meghalaya, India e-mail: [email protected] Debdatta Kandar e-mail: [email protected] A.K. Maji e-mail: [email protected] © Springer Science+Business Media Singapore 2016 N.R. Shetty et al. (eds.), Emerging Research in Computing, Information, Communication and Applications, DOI 10.1007/978-981-10-0287-8_22
243
244
Khwairakpam Amitab et al.
stored in the memory of the SAR receiver and processed to reconstruct the image of the surface [2]. The SAR platform flies along the azimuth direction at constant velocity. The azimuth is the direction parallel to the flight path. To obtain high azimuth resolution, a large antenna is needed to focus the transmitted and received echoes into a sharp pencil-like beam. SAR utilizes the forward movement of the platform to synthesize a huge antenna. The sharpness of the beam defines the azimuth resolution Eq. (1) A¼
Rk L
ð1Þ
where R is the slant range (line of sight distance), λ is the wavelength of the transmitted signal, and L is the length of the antenna. The range is the direction perpendicular to the flight path of the aircraft. By measuring the time difference between the transmitted pulse and received echo, the range of the reflecting object can be determined. Range resolution is the ability to separate two object points in the range direction. Mathematically range resolution Eq. (2) can be defined as R¼
ct 2
ð2Þ
where t is the pulse width and C is the speed of light. A smaller value of t will give a finer resolution. However, decreasing the v
Data Loading...
