Exploitable results Development and investigation of fast recursive algorithms for image processing and data reduction using segmentation The development of algorithms for image filtering, segmentation and data reduction is based on structural models of images which are considered as separate modelling of image domain and image intensity function. Attention has been paid to adaptive filtering and segmentation methods which yield better results as compared with linear or non-adaptive methods. A fast algorithm for trimmed mean filters has been developed as an example of non-linear filtering. Computational complexity of this algorithm is reduced to O(L.logL) operations per pixel as compared to O(L2.logL) when using the direct implementation, where LxL is the window size. A fast structure-adaptive algorithm for noise reduction in images was designed which allows noise to be filtered out without damaging the original image. The computational complexity of the respective fast algorithm is of order O(L2) per pixel as compared to the direct implementation complexity O(L4) per pixel when using LxL structuring element. A new structure-adaptive approach was applied to carry out binary segmentation and edge detection of radiographic images in order to detect regions of interest which enable the detection of low-contrast local objects on a non-homogenous background. A fast hierarchical algorithm for lossless data reduction in images using binary segmentation has been proposed as well. The main application area of results is the radiographic imaging for medical and technical diagnostics. The communication between the participating researchers was mostly realised through the Internet as well as during individual meeting and project workshops. A special ftp-server for information exchange between the participants via Internet has been installed. The purchased equipment (one personal computer with printer and one copier) was used to implement the co-operation agreement. Searching for OpenAIRE data... There was an error trying to search data from OpenAIRE No results available
