中文摘要：

实地获取农作物图像对于农作物长势以及病虫害进行监测具有重要作用,对此,结合脊波变换这一多尺度图像分析方法,在图像脊波变换域引入了边界判别噪声检测方法（Boundary discrimination noise detection,BDND）,对经典中值滤波算法进行了改进,提出了一种基于脊波变换域BDND改进的中值滤波算法。该方法首先对图像进行多尺度脊波变换,获得了低频和高频分解图像,考虑到低频图像的视觉特征,采用同态滤波方法进行增强处理;然后对高频图像结合区域灰度值分布特征,设定2个自适应阈值,将经过2次噪声检测后处于该2个阈值间的像素点标记为非噪声点,对其余像素点分别进行中值滤波;最后,对视觉效果改善的低频图像和滤波后的高频图像进行逆脊波变换。分别采用C＋＋语言对中值滤波、脊波域阈值去噪以及本文算法进行编程试验。结果表明,本文算法对于农作物图像的滤波效果稍优于其余2种方法。

英文摘要：

The agriculture crops images obtained in the filed play an important role in crop growth monitoring and plant diseases and insect pests monitoring. So,combing with Ridgelet transform,in the image Ridgelet transform domain,the boundary discrimination noise detection （BDNE） method is used to improve the classical median filtering algorithm,an improved median filtering algorithm based on BDND in Ridgelet transform domain was proposed. Firstly,the image was conducted multi-scale Ridgelet transform,the low-frequency and high-frequency sub-images were obtained,according to the visual features of the low-frequency sub-image,the homomorphic filtering algorithm was used to enhance the image contrast; then,according to the grey values distribution characteristics of the image regions,two adaptive thresholds were set,the pixels of grey values between the two thresholds after two noise detection were regarded as the non-noise pixels,the other pixels were possessed by median filtering algorithm; finally,the low-frequency sub-image with improved visual effects and the high-frequency sub-images after filtering were conducted inverse Rdigelet transform, The programmes of median filtering algorithm and the filtering algorithm of wavelet threshold in Ridgelet transform domain and the algorithm proposed in this paper were wrote by C＋＋ language,the ex- perimental results showed that,the performance of the proposed in this paper was better than the other two algorithms.