my_3_STEP_search

一个光流算法的matlab源程序，在图像处理、前景检测与跟踪等方面有很多应用，推荐(An optical flow algorithm matlab source code in image processing, the prospects for detection and tracking, etc. There are many applications, the recommendation)

说明：  利用遥感做变化检测的时候可以用到的数据。(Data Used in Change Detection Using Remote Sensing)

图像检索——提取颜色特征：HSV空间颜色直方图(将RGB空间转化为HSV空间并进行非等间隔量化，将三个颜色分量表示成一维矢量，再计算其直方图作为颜色特征。(Image Retrieval- extract color features: HSV space color histogram (RGB space into the HSV space and the non-interval to quantify the three color components expressed as a one-dimensional vector, then calculate the color histogram as a feature.)

该算法先通过七点进行M估计，然后挑选三点进行精确匹配运算(This algorithm firstly the fuzzy matching to achieve slightly at the beginning of localization, to achieve accurate positioning on the basis of improve the accuracy simulation)

对图像目标识别和提取，主要通过改进非线性阈值分割实现(For image target recognition and extraction, the main method is to improve the non-linear threshold segmentation.)

利用低通滤波器对图像进行插值放大：按放大的倍数对原图像进行插零，插零后的图片与低通滤波函数进行卷积，原先插入的零的位置自动替换为适当的插值，如此图像就已经插值放大了，图像质量还不错。(Double the number of pixels of each row and column by adding a zero value pixel between each pixel of the image.Then operate with two kinds of low-pass filters to get the enlarged image.)

实现图像的几何变换，用控制栅格的方法来将一幅变形的图片恢复原型(Image geometric transformation, with control of the grid method to deformation of an image to restore the prototype)

基于离散小波变换和IHS变换的遥感图像融合代码，主要用于多光谱和全色图像的融合(matlab code for remote sensing image fusion based on DWT and IHS, mainly for mutilspectral and panchromatic image fusion)

针对粒子概率假设密度(PH D)滤波算法在虚警、漏检情况下，b标状态估计不稳定和b标可观测性车弱的问题，提出了一种基于序贯融合的粒子PH D滤波方法，利用雷达和红外传感器多目标进行融合跟踪杏基本思想是先对红外传感器进行粒子PH D滤波，丙将红外传感器滤波结果作为雷达的预测值，然后利用P达观测的数据进行更新，这样通过雷达和红外传感器交替上作保证目标状态的可观测性，从滤波器输出结身即可得到目标的状态信息仿真结果表明，在虚警、漏检和密集目标环境下，该方法是有效的和稳健的(The problem of target state estimation instability and observability weakerin the presence offalse alarms and missed detection was deal with. On the basis of sequential fusion, a particleprobabil- ity hypothesis density(PHD) filter for multrsensor multrtarget tracking was proposed. Observed da to collected from the infrared sensor was estimated with the particle PHD filter. Then the results from the filter were set as the radar predicted value by the radar observations. The multrtarget state can be updated to guarantee observing the target state. In this way, the global state is updated at the fusion center. Simulation results show that the proposed algorithm is effective and robust under the false warning, omission and concentrated target environment. )

使用matlab对图像进行处理实现鱼眼校正，效果良好(Use matlab image processing fish-eye correction, and to good effect)