【作者】 王书强；

【导师】 张喜；

【作者基本信息】 北京交通大学 ， 系统工程， 2020， 博士

【摘要】 由于受到计算机技术、传感器技术、光学和数学的发展、环境场景、人为的因素等各方面的影响,有时获得的图像信息不是我们最理想的。这是由于成像设备的物理特性和图像传输时的链路特点,图像有时会对比度较差,并因不同程度的损坏而造成各种噪音或模糊不清。结合上述特点,如何从数字图像中获取更多有价值的信息,让图像切实为我们各行业服务,是图像处理领域的重要课题。数字图像的应用在医学领域和航空领域取得了巨大的成功。如今,计算机图像已经渗透到国民生产的各项活动中,每天通过各种媒体接触到的图像会给我们提供极为有价值的信息与知识。这些隐含于图像中的重要信息的获取方法引起了全世界学者们的广泛兴趣。模糊集理论提出后,显示了其在解决各种模糊性和不确定性问题强大能力。模糊集扩充了经典数学理论,形成了一个比较系统的数学分支。近五十年以来,全世界的学者在这一领域进行了各种探索和研究且取得的成果斐然,尤其是模糊理论与人工智能大数据相互结合,它的应用范围已经涉及到计算机、多媒体、自动控制、信息采集与通信等一系列高新技术行业,为推动社会进步提供了应有贡献。本文在基于模糊集理论的图像处理和控制方面展开研究,主要内容和结果如下:(1)研究了数字图像在计算机存储的的基本特征,边缘检测的定义以及每幅图像边缘生成的物理机制。提出了一种隶属度函数修正的图像增强算法。该算法首先将样本图像进行变换映射,然后根据最大模糊熵原理将目标图像划分为多个灰度层。针对不同灰度层的特性,利用修正的隶属度函数进行增强,可以有效抑制噪声,提高图像对比度,避免灰度信息丢失。在实现模糊隶属度函数准则优化的同时,尽可能多地保留图像的边缘位置和细节信息,同时确定优化参数的选取,保证增强图像的质量,提高算法的可行性和效率。从实验仿真结果可以看出,无论是从主观分析还是客观判断,实验结果都表明本文方法对图像增强是有效可行的。(2)研究了模糊集的熵测度及相关理论,利用直觉模糊集(IFS)提出了一种散度测度和熵测度的计算方法,并对其有效性进行了实验证明。对现有的熵测度公式进行了参数化整合,得到了新的散度测度和熵测度。这些参数在实际应用中具有较高的灵活性,并且参数值必须根据数据本身来确定。按照本文提出的方法对多个测试样本图像进行了测试,该方法检测出的边缘清晰平滑,并且峰值信噪比(PSNR)始终等于或大于现有的其它方法,效果良好。并将最终结果与前人的研究成果进行了比较,发现本文方法可以获得更好的结果,并且各个样本图像的检测出边缘也较为接近真实,线条较为平滑和清晰。实验表明,该方法相对于其他方法具有更好的鲁棒性和有效性。(3)研究了模糊控制及推理规则,提出了一种基于图像的模糊控制方法并用实例加以验证。在许多实际的控制系统中,受控系统需要相互协调的子系统以确保系统的正常运行。本文采用了基于图像检测模糊控制方法设计了一种模糊控制器,并进一步将控制器用于带摄像头的四轮移动机构上来实现自动物料堆放的系统,实验结果表明控制器具有一定的实用性,对基于图像处理的控制系统有很大拓展。更多还原

【Abstract】 Due to the influence of the development of computers,sensors,optics and mathematics,human factors,environmental scenes,etc.,sometimes the image information obtained is not optimal.This is owing to the physical characteristics of the imaging device and the characteristics of the link when the image is transmitted.The image sometimes has poor contrast,and various noises or blur caused by different degrees of damage.Although the image itself is stored in digital form,it is visually ambiguous.Fuzzy techniques are non-linear and knowledge-based.If such defects arise from ambiguity rather than randomness,fuzzy techniques can handle imperfect data.That being said,how to obtain more valuable information from digital images and make the images effectively serve our various industries is an important topic in the field of image processing.The application of digital images has experienced rapid development in the 1960 s with the emergence of the third generation of computers,especially in the medical field and aviation field.Nowadays,computer images have penetrated into various activities of national production,and the images exposed through various media every day will provide us with extremely valuable information and knowledge.The acquisition methods of these important information hidden in images have aroused extensive interest from scholars all over the world.Since the fuzzy set theory was proposed by Professor L.A.Zadeh of American cybernetic expert in 1965,it has shown its powerful ability in solving various problems of ambiguity and uncertainty.Fuzzy set theory expands the classical mathematics theory and forms a more systematic mathematical branch.In the past fifty years,scholars all over the world have conducted various explorations and researches in this field and achieved remarkable results,especially the combination of fuzzy theory and artificial intelligence big data.Its scope of application has involved computers,multimedia,A series of high-tech industries such as automatic control,information collection and communication have contributed to promoting social progress.The main work and results of this dissertation are as follows:(1)The basic characteristics of digital images stored in computers,the definition of edge detection and the physical mechanism of edge generation of each image are studied.An image enhancement algorithm based on membership function correction is proposed.The algorithm first transforms the sample image into domains,and then divides the target image into multiple gray levels according to the principle of maximum fuzzy entropy.According to the characteristics of different gray layers,the modified membership function is used to enhance.It can effectively suppress noise,improve image contrast,and avoid loss of gray information.While realizing the optimization of fuzzy membership function criterion,the edge position and detailed information of the image are kept as much as possible,and the selection of optimization parameters is determined to ensure the enhancement of the image quality and the feasibility and efficiency of the algorithm.From the experimental simulation results,it can be seen that whether it is subjective analysis or objective judgment,the experimental results show that the method in this paper is effective and feasible for image enhancement.(2)Using intuitionistic fuzzy sets(IFS),a new method for calculating divergence and entropy measures,namely the edge detection method of intuitionistic fuzzy divergence measure,is presented,and its effectiveness is experimentally proved.The existing entropy measurement formula is parameterized and integrated,and a new divergence measure and entropy measure depending on the order α are obtained.According to the method proposed in this paper,multiple test sample images were tested,and IFS was used to achieve the construction and application of the proposed edge detection method.The edges detected by this method are clear and smooth,and the peak signal-to-noise ratio(PSNR)is always equal to or larger than other existing methods,and the effect is good.The final results are compared with the research results of Canny,Sobel,Chaira and others,and it is found that the method in this paper can obtain better results,and the detected edges of each sample image are also closer to the real,and the lines are smoother and clearer.This shows that this method has better robustness and effectiveness than other methods.(3)The fuzzy control and fuzzy rules are studied,and an image-based fuzzy control method is proposed and verified by an example.In many practical control systems,the controlled system needs mutually coordinated subsystems to ensure the normal operation of the system.Non-linear large-scale systems are composed of many interconnected subsystems.Unlike the classic control technology,large-scale systems are not only controlled by a single controller,but also consist of independent subsystems to form a group of corresponding discrete controllers,which greatly increases the difficulty of controller design.In this paper,a fuzzy controller is designed based on the image detection fuzzy control method,and the controller is further applied to a four-wheel moving mechanism with a camera to realize an automatic material stacking system.The experimental results show that the controller has certain practicality.The control system based on image processing has been greatly expanded.更多还原