基于DSP的网络摄像机图像预处理技术

发布时间：2018-01-15 13:37

本文关键词：基于DSP的网络摄像机图像预处理技术　出处：《浙江大学》2014年硕士论文　论文类型：学位论文

【摘要】：随着硬件处理器,图像传感器性能的进步,视频监控技术飞速发展,网络摄像机大规模普及。图像质量是摄像机的核心竞争力,同时也是视频分析技术的基础。而高质量的视频图像不仅需要高性能的硬件,更需要高效的图像预处理算法配合硬件。论文对网络摄像机的预处理算法进行研究和优化。并在DM8127实验平台上进行了实现和验证。论文首先对摄像机图像采集模型进行了分析和研究。结合图像质量的影响因素,论文对图像质量的优化目标进行了讨论,并以此为指导进行预处理算法的研究优化。其次,论文对摄像机3A算法进行了研究。在自动对焦部分,论文对清晰度评价函数进行了比较,并针对自动对焦的特殊情况对自动对焦搜索方式进行了改进,在不同搜索区域使用不同步长进行搜索,提高搜索效率。在自动曝光部分,论文分析了光圈对景深模糊的影响,提出了一种图像清晰度与噪声的综合评价函数,通过寻找此函数的最值进行自动曝光的优化。此方法在硬件上实现并测试,实验结果证明此方法可以在景深模糊与噪声之间寻找到最佳的平衡。在自动白平衡部分,论文对灰度世界算法进行了改进,将图像分区域寻找灰色区域,通过灰色区域进行光照估计,并根据估计结果摄像机参数进行迭代调整。实验结果证明此算法有较好的效果。最后,论文在DM8127实验平台上对图像预处理模块进行实验和综合优化。根据场景情况,自适应地配置每个模块,实现图像质量的自动优化调节。论文基于DM8127的软件框架,将综合调节模块在硬件上实现。实验证明调节后的结果相比原始图像在主观视觉质量方面有较大程度的提高。
[Abstract]:With the progress of hardware processor and image sensor performance , video monitoring technology is rapidly developing , and the network camera is widely popularized . Image quality is the core competence of the camera , and it is the basis of video analysis technology . High quality video image not only needs high performance hardware , but also needs high - efficient image preprocessing algorithm to match hardware . In this paper , the preprocessing algorithm of the network camera is studied and optimized , and the implementation and verification are carried out on the DM8127 experimental platform . Firstly , the image acquisition model of the camera is analyzed and studied . Based on the influencing factors of image quality , the optimization goal of image quality is discussed . Secondly , the paper studies the algorithm of camera 3A . In the auto - focus part , the paper compares the definition evaluation function and improves the auto - focus search mode for the special case of auto - focus . In the auto - exposure part , the paper analyzes the influence of aperture on the blur of field depth . In the auto - exposure part , the paper analyzes the influence of aperture on the blur of field depth . Finally , the experiment and comprehensive optimization of the image preprocessing module are carried out on the DM8127 experimental platform . According to the situation of the scene , each module is adaptively configured to realize the automatic optimization of the image quality . Based on the software framework of DM8127 , the comprehensive adjusting module is realized on the hardware . The experiment proves that the adjusted result has a greater degree of improvement compared with the original image in subjective visual quality .

【学位授予单位】：浙江大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TN948.41

【参考文献】