We found a match
Your institution may have access to this item. Find your institution then sign in to continue.
- Title
Optoelectronic device based failure management using content based multispectral image retrieval and deep learning model.
- Authors
Bhukya, Raghuram; Arunsundar, B.; Tatini, Narendra Babu; Sadala, Triveni Mohan; Elshafie, Hashim; Qamar, Shamimul
- Abstract
The need for optoelectronic devices is growing right now, but the production of these devices is having a difficult time keeping up with the advancement of the machinery, instruments, and manufacturing techniques they support. Pictures from big, unlabeled image collections are commonly retrieved using a technique called content-based image retrieval (CBIR). The availability of photographs is also growing as internet growth and transmission networks increase. This study suggests brand-new image retrieval methods for multispectral images used in Optoelectronic device monitoring that are based on image segmentation and classification methods with deep learning techniques for failure management. Optoelectronic device Monitoring field-based multispectral images were used as the input, which was then processed for noise removal, resizing, and smoothing. Fuzzy c-means clustering-based image segmentation was used to divide up this processed image into its component parts. Following that, a hybrid multilayer transfer learning perception was used to classify the clustered segmented picture. The proposed technique has an accuracy of 95%, precision of 85%, recall of 75%, F-1 score of 63%, error rate of 51%, MAP of 55%; existing MRCN accuracy of 85%, precision of 75%, recall of 63%, F-1 score of 55%, error rate of 45%, MAP of 51%, OC-LBP accuracy of 89%, the precision of 79%, recall of 71%, F-1 score of 59%, error rate of 49%, MAP of 53%.
- Subjects
IMAGE retrieval; DEEP learning; CONTENT-based image retrieval; IMAGE recognition (Computer vision); MULTISPECTRAL imaging; IMAGE segmentation; OPTOELECTRONIC devices
- Publication
Optical & Quantum Electronics, 2024, Vol 56, Issue 2, p1
- ISSN
0306-8919
- Publication type
Article
- DOI
10.1007/s11082-023-05793-7