Linear and Quadratic PSO Based Color Space Conversion for Sea Target Detection

Abstract

Sea target detection is an important goal for<br><br>military purpose and navigation. A new nonlinear quadratic<br><br>transform is presented to generate target-based color space<br><br>(TCS) for sea target detection. RGB color space is converted to<br><br>TCS through a 3*3 matrix, using quadratic transformation.<br><br>PSO algorithm is used for maximum differentiation between<br><br>target and non-target clusters which are classified with FCM<br><br>clustering method. The task of PSO is to minimize the Error<br><br>value produced by FCM classifier. Experimental results show<br><br>the ability of the new method in finding sea targets in color<br><br>images. The superiority of the TCS space to the relative<br><br>previous works and other color spaces is conspicuous.