<title>Design for HMM-based SAR ATR</title>

Abstract

This paper describes progress on an Automatic Target Recognition (ATR) system for Synthetic Aperture Radar (SAR) imagery. The system is based upon a feature extraction, data ordering, and statistical modeling paradigm. Feature extraction is performed by applying image segmentation to convert the SAR imagery into one of four pixel classes. A description of a real-time image segmentation design is given. The segmented imagery is re-ordered from a two dimensional (2D) spatial representation to a sequential representation through the use of multiple Radon '&ansforms (RT). Finally, the re-ordered data is classified by target type by appling Hidden Markov Model (HMM) decoding techniques. Performance results on the MSTAR public targets database is provided.

Support subspaces method for synthetic aperture radar automatic target recognition

“…Before finding the thresholds, the image is smoothed to reduce the noise impact in the image (speckle, thermal noise of radar, etc.). We used the same method of smoothing that Kottke et al 31 used. Two steps are used to carry out smoothing: the image is down sampled four times, and then it is up sampled to instance size using linear interpolation.…”

Target Recognition Based on Directional Filter Banks and Higher-Order Neural Networks

“…The public MSTAR database has been used previously for benchmarking ATR classification performance. A number of authors have benchmarked algorithms against this database [8,13,16]. All score below 96%, with the exception of the Radon transform HMM algorithm introduced in [11] and shown in Table 3.…”

A SAR Target Classifier Using Radon Transforms and Hidden Markov Models

“…While all three of the above approaches are template-based, the system proposed by Kottke et al [4] is based on a modeling paradigm. Feature extraction is done by first segmenting the image using the independent conditional mode (ICM) technique.…”

“…In common with Kottke et al [4], we use the ICM algorithm [5], an unsupervised, multiclass segmentation method, to segment the input target chips into four target classes: bright and dim target pixels, shadow pixels, and background pixels. To mitigate the specular effects alluded to earlier, the bright and dim pixels are then merged into a single class of target pixels.…”