Development of image mosaicing techniques for photovoltaic arrays

Image mosaicing is an image processing algorithm that evaluates a panoramİc image by combaning many image or consecutive image frames of a video. Especially it widely uses in mapping, medicine, UAV sand various implemantation of computer vision. In this work, we are focused on three basic research: realizing an image mosaicing implementation, proposing a new adaptive image mosaicing algorithm and appyling the image mosaicing approach on photovoltaic arrays for tracking and controlling. In this thesis, first, the other image mosaicing algorithms in the literature have been analyzed and an image mosaicing implementation has been developed. The panoramic image is obtained by applying feature extraction, image matching, outlier point detection and removal, homography estimation and image fusion on the images taken as input, respectively. As a second step, a new image mosaicing algorithm has been proposed. This algorithm based on the number of attributes that affects performance of image mosaicing, the ratio of similarity and setting the set center continual by optimization. These three parameters have been optimized by using genetic algorithm, aiming to improve the computational complexivity and accuracy of image mosaicing, simultaneously. The efficiency of this new approach has been presented. Experimental applications of the image mocaising algorithm are made on various images and photovoltaic array images rather difficult to combine according to other images because of it has high similarity. Especially in large sized photovoltaic arrays or in solar power plants, this approach that would be used to combine image from several cameras, will get some significants on following and controlling such systems. Performance of the algorithm is verified with the experimental results and it is analyzed with comparative results. The results show the effectiveness, computational complexity and accuracy of the algorithm. Consequently, three basic implementations realizing an image mosaicing implementation, proposing a new adaptive image mosaicing algorithm and appyling the image mosaicing approach on photovoltaic arrays, have been examinedand verified with the experimental results.