基于改进HVS的水稻杂草图像分割算法  

Rice Weeds Image Segmentation Algorithm Based on Improved HVS

作　　者：李聪[1] 高海燕[2] 袁超[1] 

机构地区：[1]河南农业大学理学院,河南郑州450002 [2]华北水利水电学院数学与信息科学学院,河南郑州450011

出　　处：《计算机仿真》2012年第4期324-327,共4页Computer Simulation

摘　　要：研究水稻杂草图像分割问题,提高分割的准确性。水稻在颜色和形态上与杂草像素差异很小,造成二维图像像素重叠。传统的基于颜色和形态模型的分割算法,在这种情况下,很难准确的对混合杂草的水稻图像进行准确分割。针对分割效果不准确的问题,提出一种基于改进HVS模型的水稻杂草分割算法。通过在传统模型中加入图像亮度特性、像素频率特性、颜色的感知特性,以改进传统的水稻视觉图像模型。避免了传统分割算法对颜色和形态特征过度依赖的弊端。实验证明,利用改进后的图像视觉模型能够在重叠、间隔的水稻图像中,准确的分割杂草图像,取得了令人满意的结果。The traditional segmentation algorithm based on color and form is difficult to accurately segment the rice and weeds mixed images,and segmentation effect is not ideal.The paper put forward an improved HVS algorithm for the segmentation rice and weeds.The traditional model was joined with the characteristics of brightness,pixel frequency,and color perception,in order to improve the traditional rice image model.Experiments show that the improved image visual model can accurately segment weeds from the images and has obtained satisfactory results.

关 键 词：杂草分割 亮度特性 频率特性 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]