基于图像的摄像机焦距参数的算法研究

摘要

Fusion of foreground and background in the virtual studio needs parameters of pan, tilt and zoom. First, two parameters are collected by encoder, while the collection of focal paremeter nees complicated supporting mechanism installed in the lens, and still precise parameter can not be collected in the application. This paper dis-cusses the image processing based on the use of the improved GA algorithm and the combination of LM algorithm to realize the camera self - calibration of the fundamental matrix and the Kruppa equations optimize operations to get the camera internal parameters, in which the camera focal length is a parameter. When the internal reference value is de-termined , the camera focal length can be a parameter for the virtual studio camera focal length to provide accurate da-ta.%在虚拟演播室中前景和后景的融合需要摄像机的平摇、俯仰及变焦参数.前两个参数是利用编码器来采集,而使用编码器采集焦距参数需要额外在镜头处加装复杂的机械支撑机构,在应用中不能采集到满足精确度要求的参数值.为此探讨了基于对图像的处理,利用改进的GA算法和LM算法的结合,实现对摄像机自标定中的基本矩阵及其Kruppa方程的优化运算,从而得到摄像机的内参.由于焦距属于摄像机的一个参数,当内参值确定时,便能得到摄像机焦距参数.从而能为虚拟演播室提供精确的摄像机焦距数据.