Justyna Stypułkowska
supervisor: Przemysław Rokita
The ongoing doctoral thesis focuses on the problem of detection and classification of objects in photos recorded with a multispectral camera. The main objective of the research is, firstly, to select the most effective algorithms for detecting objects on the acquired images, and secondly, to examine the impact of the use of individual spectral channels and appropriate camera settings during image recording on the effectiveness of these algorithms. This will allow for the development and optimization of the entire process not only from the implementation side, which is a common procedure, but also from the hardware side, which will indicate the right direction of the image acquisition method in order to achieve the most effective method of detection and classification of objects recorded on them. The research uses digital image processing and analysis algorithms, artificial intelligence techniques and deep learning algorithms.
The research is innovative, because so far no one has studied the impact of the use of individual spectral channels on the effectiveness of the detection process and classification of objects in the acquired images. The results of the conducted research will allow for the proper selection of both algorithms and hardware solutions, which are key at the stage of the image acquisition itself. The developed solutions will be implemented in the work of the Remote Sensing Department of the Łukasiewicz Research Network - Institute of Aviation.