基于三维图像快速分层滤波的聚焦梯度模拟方法

A NEW METHOD OF SIMULATING FOCUSING GRADIENT BASED ON FAST STRATIFIED FILTERING ON 3D IMAGE

基于透镜成像模型推导出模糊函数σ与物距及透镜焦距的关系．采用对深度三维图像自适应分层滤波方法，引入差异度函数，按差异度函数把深度图像分解成Ｋ个ｚ方向上不相交的子图像，通过对子图像的滤波合成，降低算法的复杂度，并使滤波后的误差控制在给定范围内，引入ａｌｐｈａ值、按层合成等手段解决滤波后的遮挡及边界问题．

The relationship between fuzzy function σ and the distance from object to lens and focal length was provided by using the lens model. A new adaptive stratified filtering method for 3 D depth image was developed. The concept of the difference function was proposed and the depth image was resolved into several non intersection sub level images along z axis. The algorithm complexity was reduced and the postfiltration error was controlled within a given range by filtering and composing sub level images. The problems of boundary and obstruction were solved by introducing alpha value and composing subimages level by level. This method has been successfully used in computer graphics rendering algorithm and an example of simulating defocus of camera was given.