Data extraction with similar techniques: satellite images and medical images

This thesis study searches how common algorithms shall be successful in terms of ?MC (Micro Calcification) Extraction from Mammography images? and ?automatic capturing of buildings from satellite images? representing completely different working areas. This study identified that matters regarding above stated two different disciplines could be eliminated by means of similar algorithms under common image processing heading.In accordance with data of World Health Organization, assistance is provided to the radiology experts during diagnosis of more than one million breast cancer patients in the world through MC (Micro Calcification Extraction study. This thesis study aims to identify micro calcifications representing basic working area regarding mammography image of CAD systems and identification of its type as non-malignant and malignant. By this way, it is aimed to contribute to the accurate diagnosis of radiology experts.The objective of the study regarding automatic capture of buildings by means of satellite images is to produce an algorithm and software having the capability to create rapid and accurate results with full-automatic process. By this way, large scale of workforce, time saving and cost decrease shall be achieved during data production to provide rapid and accurate intervention in case of disasters requiring extremely rapid intervention, identification of illegal housing at metropolises and creation, as well as updating of urban models.Basic algorithms covered by the thesis study presented and created to eliminate problems include common, specific and automatic algorithms, such as utilization of mathematical morphologic operators, storage of database information for RGB color scheme, noise suppression algorithms. In addition, local threshold identification method, algorithms regarding classification of MCs, algorithms providing classification of MCs as non-malignant and malignant, creation of buildings at housing regions, identification and ordering of building corners having four or more corners are specially developed for related topics within the thesis study presenting specific and full-automatic algorithms.Software is created by programming algorithms created for both headings stated above through Microsoft Visual C# Express Edition. Algorithm and the software operate automatically and independent from human, computer and the location during elimination of related matters. Average success level of both studies is approximately 85%.