摘要
为了提高分辨率较低的毫米波隐匿武器辐射图像的可读性,研究了基于视觉显著性分析的毫米波辐射图像的增强方法。根据视觉显著性分析方法的原理,选取适合于毫米波辐射图像增强的IG模型。对图像进行频域滤波,通过设计截止频率保留目标边界并去除噪声与目标内部的纹理,在Lab颜色特征空间内计算滤波后图像的显著图作为图像增强的结果。与原图相比,增强后的图像在抑制背景的同时可以突出目标区域,并可采用自适应阈值分割的方法提取目标区域及其轮廓。采用该方法对实测毫米波隐匿武器辐射图像进行图像增强实验,证实了其增强效果在主观视觉及客观评价指标上均优于其他常用方法。
In order to improve the intelligibility of radiometric millimeter-wave ( MMW ) image of concealed weapon with low resolution, an image enhancement method based on visual saliency analysis is proposed. According to the principle of the visual saliency analysis, the model IG is selected for its suitability for MMW radiometric image enhancement. The image is filtered in frequency domain. The cutoff frequency is designed to reserve target boundaries and reduce noise and the texture in target region. The saliency map of the filtered image is calculated in Lab color feature space and considered as the enhancing result. Compared with the original image, the target region becomes prominent and the background is suppressed in the enhanced image. Moreover, the target region and outline can be obtained via adaptive threshold segmentation on the saliency map. Using the proposed method to enhance the MMW radiometric image of concealed weapon,the experimental result demonstrates better performances in both respects of subjective vision and objective criteria than the common methods.
出处
《南京理工大学学报》
EI
CAS
CSCD
北大核心
2014年第1期134-139,共6页
Journal of Nanjing University of Science and Technology
基金
国家自然科学基金(61301213)
高等学校博士学科点专项科研基金(20093219120018)
关键词
隐匿违禁物品探测
毫米波辐射图像
图像增强
视觉显著性
自适应阈值分割
concealed contraband detection
millimeter-wave radiometric image
image enhancement
visual saliency
adaptive threshold segmentation