SATELLITE REMOTE SENSING IMAGE BASED AIRCRAFT RECOGNITION USING WAVELET AND CURVELET TRANSFORMS

Abstract: This paper deals with the accurate recognition of aircrafts in satellite images using feature extraction and classification algorithms. Automatic recognition of an aircraft is a tough task. In previous recognition method, direction estimation technique was used to bring the aircraft into same direction, and the recognition was done by converting the recognition process into reconstruction process. Jigsaw algorithm was used in this reconstruction process. The major drawback of this method is that it has less accuracy while in recognition due to it's pixel to pixel matching process it also affected by noise. In order to improve the accuracy, noise components was removed by means of PCA algorihm, DWT and FDCT techniques were used for segmentation. Then the feature values of segmented images are get extracted using GLCM matrix. Recoganition of aircraft was done by comparing the feature values of input images with data base images using Probablistic Nueral Network (PNN), finally performance metrics was analyzed with the help of OTSU method. Experiments were done on more number images, it reveals that the suggested method got 96.15% of accuracy.

Keywords: Aircraft recognition, PCA algoritm, Feature extraxtion, Segmentation, Classification algorithm.