chuhuaxihua1

：植物种类识别方法主要是根据叶片低维特征进行自动化鉴定。然而，低维特征不能全面描述叶片信息，识别准确率低，本文提 出一种基于多特征降维的植物叶片识别方法。首先通过数字图像处理技术对植物叶片彩色样本图像进行预处理，获得去除颜色、虫洞、 叶柄和背景的叶片二值图像、灰度图像和纹理图像。然后对二值图像提取几何特征和结构特征，对灰度图像提取 Hu不变矩特征、灰 度共生矩阵特征、局部二值模式特征和 Gabor 特征，对纹理图像提取分形维数，共得到 2183 维特征参数。再采用主成分分析与线性 评判分析相结合的方法对叶片多特征进行特征降维，将叶片高维特征数据降到低维空间。使用降维后的训练样本特征数据对支持向量 机分类器进行训练(plant species identification method is mainly based on blade automatic identification of low dimensional characteristics.However, can not fully describe blade low-dimensional feature information, identification accuracy is low, in this paper A kind of plant leaves recognition method based on multiple feature dimension reduction.First by digital image processing technology to the plant leaf color sample image preprocessing, obtain background color removal, wormhole, petioles, and the blades of a binary image, gray image and texture image.Then the binary image to extract the geometric characteristics and characteristics of structure and characteristics of gray image extraction Hu moment invariants, gray co-occurrence matrix feature, local binary pattern features and Gabor, to extract the fractal dimension of texture image, get 2183 d characteristic parameters.By principal component analysis and linear uation analysis method of combining the characteristics of blade more feature dimensi)

中科院图像理解与计算机视觉课件6 值得参考一下 some topics on pattern recognition and computer vision(Chinese Academy of Sciences and Computer Vision Image Understanding Courseware 6 worth refer to some topics on pattern recognition and computer vision)

说明：  最二小乘影像匹配，精度可以达到0.01个像素，也可以最二小乘影像模板匹配。经过修改后可以直接用，只要调用一个函数就搞定，函数参数的说明很清楚，绝对不会有任何困难，非常好用。(Most two Theravada image matching accuracy can reach 0.01 pixel images can also be the most two Theravada template matching. After modifications can be directly used, as long as is you call a function, function parameters, the description is clear and will not have any difficulties, very easy to use.)

说明：  使用gpu、cpu并行进行sift算子计算匹配，能够在原来的基础上加速处理，但对显卡要求较高，具体环境配置使用方法可以参照mannual(SiftGPU is an implementation of SIFT [1] for GPU. SiftGPU processes pixels parallely to build Gaussian pyramids and detect DoG Keypoints. Based on GPU list generation[3], SiftGPU then uses a GPU/CPU mixed method to efficiently build compact keypoint lists. Finally keypoints are processed parallely to get their orientations and descriptors. SiftGPU is inspired by Andrea Vedaldi s sift++[2] and Sudipta N Sinha et al s GPU-SIFT[4] . Many parameters of sift++ ( for example, number of octaves, number of DOG levels, edge threshold, etc) are also available in SiftGPU. The shader programs are dynamically generated according to the parameters that user specified. SiftGPU also includes a GPU exhaustive/guided sift matcher SiftMatchGPU. It basically multiplies the descriptor matrix on GPU and find closest feature matches on GPU. Both GLSL and CUDA implementations are provided. )

说明：  对灰度图象进行正交变换,并进行滤波,产生平滑图象(Gray-scale images of the orthogonal transformation, and filtering to produce a smooth image)

this is a project about animation. how do we form the shape to make it look like car, house abd so on. with the colours that wee like and the movement.

IMM交互式对机动目标跟踪，程序写的很完美，包含各种常见的滤波算法(IMM UKF KF)

利用CSplitterWnd类实现对单文档工程进行视图区域的分割(Using the CSplitterWnd class implements the single document engineering for segmenting the view area)

对lagrange函数公式进行代码化，可以对一组数进行插值算法，可以输出最后的y值(The lagrange function formula code, a number of groups interpolation algorithm, you can output the final value of y)

C++和halcon结合，用于图形处理时进行三相机标定(C++ and halcon combined for three camera calibration)