DETECTION AND CORRECTION OF DISTORTED IMAGES OF MEDICAL OBJECTS BASED ON CONVOLUTIONAL NEURAL NETWORKS

R. Safarov

Samarqand davlat universiteti ilmiy tadqiqotlar axborotnomasi · 2026-yil

Annotatsiya

Scientific and methodological foundations for optimal identification, recognition, and classification of single-cell medical objects have been developed using hybrid convolutional neural networks with recurrent learning, as well as image transformation and filtering mechanisms. Algorithms for training networks with various architectures have been investigated. Adapted neural network architectures with independent recurrent layers have been proposed. Mechanisms for detection, localization of selected areas with distorted points, and error correction using the Daubechies wavelet transform 4, 7 were implemented. Support vector machines and gradient optimization mechanisms were proposed for network training. A software package has been developed for visualization, recognition, and classification of images of medical objects based on the use of standard search, recognition, and classification tools, and CUDA parallel computing technology.

Maqola ma’lumotlari
MualliflarR. Safarov
JurnalSamarqand davlat universiteti ilmiy tadqiqotlar axborotnomasi
Nashr sanasi2026-07-01
Jild1
Son3
Betlar111-119
DOI10.59251/2181-3973.2025.v1.138.1.3949

Samarqand davlat universiteti ilmiy tadqiqotlar axborotnomasi jurnalidan boshqa maqolalar

Samarqand davlat universiteti ilmiy tadqiqotlar axborotnomasi — barcha maqolalar