Genetic programming for automatic target classification and recognition in synthetic aperture radar imagery

Abstract

Abstract. We use the genetic programming (GP) paradigm for two tasks. The first task given a GP is the generation of rules for the target / clutter classification of a set of synthetic aperture radar (SAR) images, the second, the generation of rules for the identification of tanks in a second set of SAR images. To perform these tasks, previously defined feature sets are generated on the various images, and GP is used to select relevant features and methods of analyzing these features. GP results are then compared with previous work using the feature sets.

Learning Features for Object Recognition

“…Roberts and Howard [4] use GP to develop automatic object detectors in infrared images. Stanhope and Daida [5] use GP for the generation of rules for target/clutter classification and rules for the identification of objects. Unlike the work of Stanhope and Daida [5], the primitive operators in this paper are not logical operators, but operators that work on real numbers.…”

“…Stanhope and Daida [5] use GP for the generation of rules for target/clutter classification and rules for the identification of objects. Unlike the work of Stanhope and Daida [5], the primitive operators in this paper are not logical operators, but operators that work on real numbers. They use GP to evolve logical expressions and the final outcome of the logical expressions determines the type of object under consideration (for example, 1 means target and 0 means clutter); we use CGP to evolve composite feature vectors for a Bayesian classifier and each sub-population is responsible for evolving a specific composite feature in the composite feature vector.…”

Automatic Design and Synthesis of Automatic Target Recognition (ATR) Systems Using Learning Paradigms

“…Their initial experimental results showed that GP is a viable way of synthesizing composite operators from primitive operations for object detection. Stanhope and Daida [8] used GP to generate rules for target/clutter classification and rules for the identification of objects. To perform these tasks, previously defined feature sets are generated on various images and GP is used to select…”

“…The white region in Figure 2.11(b) corresponds to the river to be extracted. The training regions are from (68,31) to (126, 103) and from (2,8) to (28,74). The testing SAR image is shown in Figure 2.14(a).…”

Morphological algorithm design for binary images using genetic programming

“…Indeed, GP has provided excellent results in programming environments, such as parallel algorithms [32,23] and quantum algorithm [36], where human programmers struggle. Furthermore, GP has been applied successfully to solve other difficult problems in the image analysis domain such as classification [40,37,46], detection [38,2,43,16,31], recognition [41,1,47], segmentation [29], analysis [22,24], compression [21], feature construction [30] and stereo vision [11]. However, to date very limited work has been done on using GP in the important area of mathematical morphology algorithm design.…”