关键词: 高分辨距离像, 自动目标识别, 等间隔划分角域, 自适应划分角域, 自适应高斯分类器, 高斯过程分类器

Abstract:

In the field of high resolution range profile (HRRP) based radar automatic target recognition (ATR), it is an effective approach to tackle the target-aspect sensitivity by segmenting continuous radar data into several sectors in equal angular intervals. This equal angular-sector segmentation algorithm is based on the condition for good quality of ISAR images, but it is not a satisfactory choice with the application of ATR for its inclination to cause model mismatch, performance limitation, and time-consumption. This paper proposes a recursive algorithm for adaptively angular sector segmenting based on Adaptive Gaussian Classifier (AGC) and Gaussian Processes classifier (GPC).Since the HRRP data are continuous along the azimuth,we first exploit the nonlinear structure characteristic embedded in HRRP data through AGC or GPC, then present a criterion for determining the angular-sector boundary, and finally recursively segment the data. Promising experimental results are presented for measured radar data.

Key words: high resolution rang profile (HRRP), automatic target recognition (ATR), equal angular-sector segmentation, adaptive angular-sector segmentation, Adaptive Gaussian Classifier (AGC), Gaussian Processes classifier (GPC)