高斯混合模型在海面运动目标检测中的应用

Application of Gaussian Mixture Model in Moving Target Detection of Sea Surface

针对海面背景图像的特点，利用高斯混合模型参数估计方法，建立了象素点背景模型并用于海面运动目标的检测。提出的基于变化检测的高斯混合模型参数估计方法与EM算法估计的模型参数极为相似，但前者大大节省了建立背景模型所需的内存空间，建模速度大约是后者的两倍。视频流实验验证了算法的有效性。

This article proposes a parameter estimation method for Gaussian mixture model based on change detection, and uses the method to create background model for each pixel for the characteristic of sea surface. The proposed method has the similar property with EM algorithm. But the proposed method has better property in compute memory space and speed. The result of video stream ships detection experiment is satisfactory