Development of real-time shadow detection algorithms for reconfiguration in PV arrays

In todays, computer vision based algorithms is used in many fields such as in solving real-world problems with the progress of image processing technology. Especially, the use of image processing algorithms in industrial problems provide the development of low measure and high-precision solution with lower computing costs and sensorless approaches. One of the factors that affect performance such as object detection with image processing techniques, object tracking, edge detection and anomaly detection is shadow. Therefore, various applications in the literature are studied about the detection of the shadow. The most important one is shadow detection algorithms that developed to simplify distinguish from the shadows to objects in order to detect moving vehicles. As well as, shadow detection is a complex problem in terms of image processing, variety of approaches are used according to applications. Chromacity-based algorithms are based on chromacity values of the pixels in the images, physically based algorithms that use the physical properties of the image, geometric-based algorithms that uses geometric properties of objects in the scene and pixel communities grouped similarities in more than one property are classified as texture-based algorithms. In the first place, in this thesis comparison results were obtained through the experimental data in order to demonstrate in different applications performances of shadow detection algorithms via creating a shadow model. After then, new shadow detection algorithms suitable for used in real-time has been proposed according to the information obtained from this comparison results. In the third part constitutes the main aim of the thesis, for shadow detection has a very important place of the photovoltaic(PV) array, input information was obtained re-configuring approach in the PV array by using algorithms developed within the thesis. Being effectively at PV orders, determining shadow and the intensity of shadow fort he shadow determining algorithms which is used to find the best link configuration that is going to give maximum energy, artificial immunity system based and fuzzy logic based of shadow detecting algorithm had been suggested and executed as real time. Proposed shadow detection algorithms has been tested on many image with validated experimental data.The extent of thesis, developed activities of shadow detection algorithms has been demonstrated with several shadowing many scenarios in order to find needed especially in terms of re-configuring in the PV array shading values where the images taken from the camera. When the obtained experimental results taken into account, proposed real time shadow detection algorithms have been shown to be high accuracy,low-complexity and effective. Especially the a results of the determination place and intensity of partial shading that occurred in the PV array have obtained the highest power connection scheme according to current status of PV array and so has productivity gains in significant ratio.Although there are partial disadvantages as the point in object sections are similar to noise and shadow in image of application detection algorithms.Algorithms provide effective results in the scenarios which consist of clean of noise, it is known in the image of the object model and the dissimilar objects region to shadow region. As a result at this thesis work, the qualifications and model of shadows had been presented, image entering based of shadow determining algorithm had been analized relatively, artifical immunity and dreamy logic based algorithms had been offered and with an aim for a new order at PV array, using image entering based shadow determining algorithm, important advantages had been presented. All of the studies and proposed methods validated with various scenarios on experimental data and carried out in real time. The studies aimed to develop algorithms in this thesis is supported by TUBITAK 112E214 No. 1001 research project.