Tiling algorithm with line-based transform for rapid ship detection and wake feature extraction in ALOS-2 SAR sensor data

Tiling algorithm with line-based transform for rapid ship detection and wake feature extraction in ALOS-2 SAR sensor data

The satellite PALSAR-2/ALOS-2 image plays a vital role in marine environment protection, weather analysis, and other geophysical processes. Satellite image processing is requiring more time to extract the different features in the ocean region. In this article, the proposed tile-based iteration technique along with line transform for high-speed ship detection and wakes feature extraction in the PALSAR-2/ALOS-2 image. The standard noise removal/despeckling techniques are applied to the SAR data to get accurate information about the ship for detection and wake features for velocity estimation. The despeckling performance is evaluated by the metrics called Edge Preservation Index (EPI) and Structure Similarity Index (SSIM) metric. Furthermore, the proposed tiling and region-based morphological operation are applied to the despeckled SAR image for ship detection and the results are compared with an exciting method. Line-based Radon and Hough transform has estimated the angle and wavelength of the ship wake lines which are present behind the ship. The article is focused on rapid ship detection and feature extraction which consumes only 160 seconds to process the tile-based combination of image size 1978 × 2536, 7 times faster than without tilling algorithm. To demonstrate the proposed work, the test environment is created using MATLAB and the result proves that 1.0x to 1.5x faster than the standard existing method.