An Innovative Remote Sensing Image Retrieval Techniques Based on Haar Wavelet-LTRP and ANFIS

In the existing Remote sensing image retrieval methods, the images are extracted from the remote sensing image database by means of three characteristic models they are visual features, object features and the scene feature. Although this technique achieves high retrieval accurateness, the precision value is low down. In order to enhance the efficiency and precision value even more higher a Haar wavelet based LTRP technique is proposed in this paper. To extract the object feature instead of earlier new watershed segmentation(NWS) method, Haar wavelet based LTRP technique is used. In the proposed technique, at first the visual features are removed from the images by means of the spatial spectral heterogeneity technique. Afterwards the object features are removed by applying the Haar wavelet and the object features are extracted from the wavelet band by developing LTRP method and the Extracted object features are classified by Neuro-Fuzzy system typically known as ANFIS. Subsequently Scene semantic models are used for the recovery of parallel scene images from the database. The projected RSIR system based on Haar wavelet-LTRP and ANFIS technique are executed in working platform of MATLAB. The performance is measured by utilizing a compilation of remote sensing images taken from the database. In addition the recital is examined by comparing the projected Haar wavelet-LTRP method with the NWSRSIR and the usual SBRSIR method in terms of performance estimate metrics such as precision, recall and F-Measure rate. The implementation effects show the competence of projected Haar wavelet based LTRP method in Remote sensing image retrieval method.