基于SRIO(SerialRapidIO)多模型目标检测跟踪系统设计与研究

Design and research of multi-model target detection and tracking based on SRIO(Serial RapidIO)

技术的飞速发展具有对源自光电传感器的数字图像进行目标识别与跟踪为核心技术的光电成像跟踪设备在人工智能、信息安全、自动控制、计算机视觉、模式识别、智能交通、视频监控等领域具有越来越广泛的应用。本文对图像背景和目标成像特性复杂多变的特点进行了分析,提出了多模型目标检测跟踪结构研究,以扩展目标的检测跟踪为例,采用将差分检测结果与边缘检测结果融合的方法,有效抑制了分割噪声与虚警,并进一步采用多特征方法结合先验知识,提高了检测效率。在目标跟踪环节,通过将对比度跟踪和相关跟踪结合的方法,建立了粗精两级跟踪的框架,有效提高了跟踪精度,避免了单一算法跟踪精度较低的缺点。在对比度粗跟踪过程中,采用多特征关联方法提高了识别的可靠性,通过对应的仿真实验分别验证了多模检测与跟踪方法的有效性。

With the rapid development of modern photoelectric sensor technology and computer technology,photoelectric imaging tracking equipment derived from photoelectric sensors has become more and more widely used in the fields of artificial intelligence,information security,automatic control,computer vision,pattern recognition,intelligent transportation,and video monitoring.In this paper,the complex and changeable characteristics of image background and target imaging are analyzed,and the multi-model target detection and tracking structure is studied.Taking extended target detection and tracking as an example,the method of fusion of difference detection results and edge detection results is adopted.The segmentation noise and false alarm are effectively suppressed,and the detection efficiency is improved by multi-feature method combined with prior knowledge.By combining contrast tracking and correlation tracking,the framework of two levels tracking is established,which effectively improves the tracking precision and avoids the disadvantages of the lower tracking precision of a single algorithm.In contrast coarse tracking,the reliability of recognition is improved by using multi-feature association method,and the effectiveness of multi-mode detection and tracking method is verified by corresponding simulation experiments.