多尺度加权图像块局部对比度的红外小目标检测  

Infrared small target detection based on multi-scale weighted local contrast

作　　者：张丹 王军锋[1] ZHANG Dan;WANG Jun-feng(School of Science,Xi’an University of Technology,Xi’an 710054,China)

机构地区：[1]西安理工大学理学院,陕西西安710054

出　　处：《计算机工程与设计》2023年第5期1459-1470,共12页Computer Engineering and Design

基　　金：国家自然科学基金项目(F060307)。

摘　　要：为解决传统算法在强杂波背景下,容易造成误检与漏检等问题,从图像块的角度提出一种局部对比度度量方法。提出中心区域与背景区域之间对比度的新度量,在计算对比度时采取行处理,简化算法的复杂性;引入可信度值作为权值,加权处理局部对比度,对加权后的局部对比度进行多尺度处理,生成目标显著图;对显著图进行阈值分割,得到真实目标。通过在实验中的分析对比,验证了该算法的有效性。The traditional algorithm in the background of strong clutter is easy to cause errors and omissions.To solve the problem,a local contrast measurement method was proposed from the perspective of image blocks.A new measure of the contrast between the center area and the background area was proposed,which was adopted to simplify the complexity of the algorithm.The local contrast was weighted by introducing the confidence value as the weight,and the weighted local contrast was processed by multi-scale processing to generate the target saliency map.The real target was obtained by threshold segmentation of the saliency graph.The effectiveness of the algorithm was verified by the analysis and comparison in the experiment.

关 键 词：红外小目标 多尺度 显著图 局部对比度 视觉注意机制 目标增强 背景抑制因子 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]