Sea and land scene segmentation method based on initial phase-Doppler invariant distance and knn

Abstract

The invention discloses a sea and land scene segmentation method based on initial phase-Doppler invariant distance and KNN, the steps of which are: transform the obtained echo sequence into a holographic clutter image; Select the sea clutter and ground clutter tags in the clutter image; select the range-wave position resolution unit; calculate the initial phase-Doppler invariant distance between each selected range-wave position resolution unit and all tags; use the K closest The adjacent KNN method determines the category of each selected range-wave position resolution unit and then obtains a binary image; performs morphological filtering on the binary image to obtain the final result. The invention calculates the initial phase-Doppler invariant distance between each range-wave position resolution unit and each label, and combines the K-nearest neighbor KNN method to obtain sea and land noise scene segmentation images.

Description

technical field [0001] The invention belongs to the technical field of signal and information processing, and further relates to a sea and land scene segmentation method based on initial phase-Doppler invariant distance and K-Nearest Neighbor KNN (k-Nearest Neighbor) in the technical field of radar signal processing. The present invention can be used for the echo data acquired by shore-based and airborne radars to realize the segmentation of the ocean area and the land-island reef area of ​​the detection scene. Background technique [0002] The target detection technology in the background of sea clutter is a crucial research direction in radar application technology, and it is widely used in military and civilian fields. Before performing object detection in the background of sea clutter, preprocessing of sea and land clutter scene segmentation is necessary. The ground clutter and island reef echoes are separated from the radar clutter through the sea and land clutter scen...

AI Technical Summary

Problems solved by technology

The disadvantage of this method is that when the features of sea and land images are extracted and fused to obtain a comprehensive feature map, the comprehensive feature map only includes texture features and grayscale features. When the gray value is similar, it is difficult to distinguish the sea area and the land area from the texture feature and the gray feature

The disadvantage of this method is that when the algorithm of the maximum between-class variance method OTSU threshold method is implemented, more manual intervention is required in the threshold setting, the robustness of the algorithm is limited, and the method is still relatively sensitive to noise, resulting in The robustness of this method is poor, and the image segmentation results are uneven

Images