基于SACON模型和五帧差分法的目标检测算法

在充分研究现有运动目标检测算法的基础上,针对当前常用运动目标检测算法易受光照和噪声的影响,不易提取完整运动目标,提出了一种新的结合SACON背景模型与五帧差分法的运动目标检测算法.对传统的SACON算法进行改进得到运动区域,与五帧差分算法提取的运动目标相结合,采用动态阈值以适应光线突变,通过孔洞填充等后处理,综合得到运动前景图像.该算法有效地处理了孔洞和噪声问题,具有很好的实时性以及抗干扰能力,能够精确地检测出运动目标.

Object Detection Algorithm Based on SACON Model and Five-Frame Difference

On the basis of fully studying the existing target motion detection algorithm,to solve the problems in current moving object detecting algorithm,such as likely to be influenced by illumination and noise,and difficult to extract the complete moving object,a new moving object detection algorithm combining SACON background model and five-frame difference method is proposed.The moving region is obtained by improving the traditional SACON method,combining with the moving object extracted by five-frame difference algorithm,using the dynamic threshold to adapt to sudden change of light ray,by means of post-processing such as pin-hole filling,the moving foreground image is achieved by integration.This algorithm effectively processes pin-hole and noise issues with better real-time property and anti-jamming capability,which can accurately detect moving object.