ICSP2008 Proceedings
Applicability Analysis of CFAR Algorithm for Landmines Detection
Applied to Vehicle-Mounted Forward-Looking SAR
CHU Ning, YANG Yan-guang, ZHOU Zhi-min
(School of Electronic Science and Engineering, National University of Defense
Technology, Changsha 410073,China )
Chuning1983@gmail.com
Abstract
2. Cluster distribution.
For the vehicle-mounted forward-looking synthetic aperture radar (SAR), detecting algorithms should be effective, robust and real-time. This paper proposed a fast statistic detecting algorithm based on circulatory iteration. We also presented feasible methods to select the optimal parameters of constant false alarm rate (CFAR) detector. Experiment results showed that our algorithm had high detecting rate, low false alarm rate, good adaptability and much less computation load.
To improve the efficiency of CFAR detector, the background cluster distribution should be analyzed firstly. For the single-look image with envelope attitude, the background clusters are more suitable to the Raylei distribution. For the noncoherent multi-look images, after the 2-D mean filtering, clusters are close to the Gauss distribution.
3. Fast CFAR algorithm.
Herein, a fast statistic detecting algorithm based on circulatory iteration is proposed.

1. Introduction.
The landmine crisis all over the world results in serious social and economical problems
[1,2].Vehicle-mounted forward-looking SAR is
3.1. Method steps
able to detect a large scale of area effectively in a short time [3,4]. Its detecting algorithms should be:
Adaptability to various environments High detectable and low false alarm rates. Concise operations
Generally, there are two ways[1-5] of detecting the SAR image.
Image separation based on the integral Detection based on the adaptive thresholds. One effective method is the CFAR detector. It utilizes the statistics of the background clusters to obtain the local threshold adaptively;
Figure 1. Window description and iterating method of
fast CFAR detector
threshold.
During the statistic procedure, CFAR window would slide orderly from the first pixel to the last one, as the figure 1 shown.
The symbols r and
______________________________________ 978-1-4244-2179-4/08/$25.00 ©2008 IEEE
g represent the
2521
热门文档