频域图像增强混合滤波算法及在作物缺素中的应用  

Mixed filter algorithm of image enhancement in frequency domain and its application in research of corps' nutrient deficiency

作　　者：杨捷[1] 冯兵[1] 吴素萍[1] 

机构地区：[1]宁夏大学数学计算机学院,宁夏银川750021

出　　处：《计算机工程与设计》2014年第9期3156-3162,共7页Computer Engineering and Design

基　　金：国家科技支撑计划课题基金项目(2014BAD05B02);国家自然科学基金项目(60963004)

摘　　要：针对目前常用的频域图像增强方法存在振铃现象、模糊效应与过度锐化,为突出图像的细节及边缘信息,保护图像原有的光滑区域灰度,提出一种新方法。利用低通和高通滤波器它们各自的优势,有选择性地对图像进行滤波处理,即对图像的高频和低频部分分别进行不同的系数增强。实验结果表明,该算法保持了图像的光滑区域灰度,改善了边缘区域的对比度,验证了该算法的有效性。针对作物缺素问题,运用不同滤波算法对采样图像进行处理后,对病斑部位采用相同的形状特征提取方法进行实验,实验结果表明,该算法突出了叶面纹络与褐色斑块,保持了样本图像的光滑区域灰度,有效改善了对病斑部位形状特征参数的提取效果,具有一定的可应用性。Resent methods for the image enhancement in the frequency domain have problems like ringing, blurring effect and excessive sharpening. Aiming at sowing these problem and highlighting the details and edge information, meanwhile protecting the original grayscale smooth area of the image, an algorithm was introduced that took advantage of both low-pass and high-pass filter, i. e. , using different coefficients to enhance the high-frequency and low-frequency of images. Experimental results justify this algorithm＇s high efficiency by maintaining the gradation of smooth region, improving the contrast of edge region, and avoid problems arising when either low-pass or high-pass filter was used. The experiment was carried out in the research of corps＇ nutrient deficiency where the same feature extraction method was taken on necrosis region after processing different filter algorithms on sampled images. By contrasting the experimental results, this algorithm shows its higher efficiency by highlighting leaf texture and brown patches, maintaining the smooth region＇s gray level on sampled images, and facilitating the extraction of disease region＇s shape feature parameters.

关 键 词：频域 图像增强 低通滤波 高通滤波 缺素 

分 类 号：TP312[自动化与计算机技术—计算机软件与理论]