On-board ship targets detection method based on multi-scale salience enhancement for remote sensing image

A on-board ship targets detection method based on multi-scale salience enhancement is proposed. Unlike the traditional wavelet filter enhancement methods, the proposed utilizes the wavelet decomposition to obtain the high-low frequency parts, and estimate the salience feature with both parts, which enhance the ship targets efficiently. First, decompose the remote sensing image by 2-D DWT, and obtain the low-frequency part, high-frequency part of horizontal, vertical and diagonal; then, compute the OSTU threshold, which is subtracted by the low-frequency coefficients to get the low-frequency salience image; and, the high-frequency parts are used to compose the high-frequency salience image; finally, the high-low parts are fused by addition and normalized to obtain the salience map. The original data of multi sets of remote sensing images are experimented, and the results are compared with the method without the proposed salience enhancement. The proposed shows obvious salience enhancement for the low-resolution, high-noise remote sensing images.