CMStatistics 2017: Start Registration
View Submission - CMStatistics
B1315
Title: Finding conics in noisy images Authors:  Agustin Mayo-Iscar - Universidad de Valladolid (Spain) [presenting]
Luis Angel Garcia-Escudero - Universidad de Valladolid (Spain)
Alfonso Gordaliza - Universidad de Valladolid (Spain)
Abstract: Trimming is a useful tool for robustifying different statistical methodologies. For the joint location-scatter estimation and the linear regression model estimation, MCD and LTS are well known trimming proposals. Now, we are interested in applying trimming for finding conics in noisy images. In order to do it, we have designed an algorithms, for different situations based on the robustification of the corresponding procrustes approaches. As usual, when applying this technique, the level of trimming will be an input parameter which should be provided in advance. We have got empirical evidences about how the methodology works obtained after applying these procedures both to artificial data and real data.