出版社：Samarskii Natsional'nyi Issledovatel'skii Universitet imeni Akademika S.P. Koroleva,Samara National Research University
摘要：As a part of the general problem of automatic information feature construction, we considered the particular applied problem of the calculation direction adjustment for the directed texture features designed to further diagnosis of various diseases by the digital biomedical images. As feature space quality criteria we considered the classification accuracy, Bhattacharyya distance and the discriminant analysis criteria. We used random search, genetic algorithm and simulated annealing as the optimization algorithms. Proposed approach reduces error probability estimation twice for the bone tissue X-ray images diagnosis problem (from 0.20 to 0.10), and also for lungs CT images diagnosis problem – by 45 % (from 0.11 to 0.06) in comparison with using of the conventional procedures for the selection of a large number of the heterogeneous features.
