RD

一个改进的自适应canny算法，该方法采用改进的开关型中值滤波器对图像做去噪工作，再把整幅图像分割为若干子图像，最后在充分考虑全局以及各子图像的边缘梯度信息后，自适应地生成动态阈值。(The new Canny adaptive method divides image into sub-images,then determinates the thresholds automatically according to the whole image edge information and the sub-images edge information.)

基于非局部图像去噪模型的遥感图像融合的变分方法。该变分方法采用梯度下降法。(Variational method of remote sensing images nonlocal image denoising model fusion. The variational method using gradient descent.)

C# 图片推拉效果 实例

改进的水平集方法，采用C_V模型进化。为多相位(improve the level set method, C_V evolutionary model. For the multi-phase)

圆形，孔型工件的尺寸测量，主要采用canny算子检测边缘，并进行亚像素细分，还有GUI，测量精度在0.1mm(Circular hole workpiece size measurement using canny operator edge detection and subpixel, as well as GUI, measurement accuracy of 0.1mm)

matlab实现的直线检测方法，主要为hough变换，包括图片及实验结果(Matlab implementation of the line detection method, mainly for the Hough transform, including pictures and experimental results)

人耳识别，包含图像预处理，图像特征点的提取，比较实用(Ear recognition)

figure 4.2a single pluse 3-D uncertainty plot pulsewidth is 2 seconds。 Figure 4.2b. Contour plot corresponding to Fig. 4.2a. Figure 4.2c. Single pulse 3-D ambiguity plot. Pulsewidth is 2 seconds. Figure 4.2d. Contour plot corresponding to Fig. 4.2c. (figure 4.2a single pluse 3-D uncertainty plot pulsewidth is 2 seconds. Figure 4.2b. Contour plot corresponding to Fig. 4.2a. Figure 4.2c. Single pulse 3-D ambiguity plot. Pulsewidth is 2 seconds. Figure 4.2d. Contour plot corresponding to Fig. 4.2c.)

图像纹理特征提取，包括分形维数，灰度游程长度，灰度共生矩阵等五种图像纹理特征。(To extract image texture features, including fractal dimension, length of gray-level run-length, gray level co-occurrence matrix and other five kinds of image texture features。)

说明：  利用稀疏表示 omp k-svd训练字典的方法，实现了图像去噪(The sparse representation of OMP k-svd training dictionary is used to realize Image Denoising)