基于帧差和背景建模的卫星视频目标检测

传统单一的目标检测方法运用在卫星视频时不能得到理想的检测效果,特别针对一些目标纹理特征匮乏,背景变化明显的卫星视频来说适用性更差。文章提出一种结合背景模型更新的混合高斯背景建模和多帧差分的算法,通过抽取吉林一号卫星视频中某2帧的2个不同场景,对比了多帧差分法、单高斯背景建模法和本文提出的算法,较好地解决了卫星视频微小目标检测中易产生检测率低、虚警率高的问题,提高了检测的正确度、完整度和质量,在遥感卫星视频运动目标检测中具有良好的应用潜力。

Target Detection Method Based on Frame Difference and Background Modeling in Satellite Video

Traditional and single target detection method can not get the ideal detection effect when it is used in satellitevideo, especially for some targets lack of texture features, and the adaptability of the satellite video with obvious back.ground changes is even worse. In this paper, an algorithm based on hybrid Gauss background modeling and multi frame dif.ference, which combined the background model updating, was proposed. By extracting two different scenes of two framesin Jilin-1 satellite video, the multi frame difference method, the single Gauss background modeling method and the algo.rithm proposed in this paper were compared. The problem of low detection rate and high false alarm rate in the process ofdetecting small targets in satellite video was better solved. Meanwhile, the accuracy, integrity and quality of the detectionwere improved. It has a good application potential in the field of moving target detection in remote sensing satellite video.