• Korean Journal of Optics and Photonics
• /
• v.32 no.6
• /
• pp.286-295
• /
• 2021

A modified catadioptric omnidirectional optical system (MCOOS) using an RGB/NIR CMOS sensor is optically designed for a capsule endoscope with the front field of view (FOV) in visible light (RGB) and side FOV in visible and near-infrared (NIR) light. The front image is captured by the front imaging lens system of the MCOOS, which consists of an additional three lenses arranged behind the secondary mirror of the catadioptric omnidirectional optical system (COOS) and the imaging lens system of the COOS. The side image is properly formed by the COOS. The Nyquist frequencies of the sensor in the RGB and NIR spectra are 90 lp/mm and 180 lp/mm, respectively. The overall length of 12 mm, F-number of 3.5, and two half-angles of front and side half FOV of 70° and 50°-120° of the MCOOS are determined by the design specifications. As a result, a spatial frequency of 154 lp/mm at a modulation transfer function (MTF) of 0.3, a depth of focus (DOF) of -0.051-+0.052 mm, and a cumulative probability of tolerance (CPT) of 99% are obtained from the COOS. Also, the spatial frequency at MTF of 170 lp/mm, DOF of -0.035-0.051 mm, and CPT of 99.9% are attained from the front-imaging lens system of the optimized MCOOS.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.17 no.6
• /
• pp.1516-1529
• /
• 2023

In this paper, a motor gearbox fault-detection system based on a hierarchical flow-based model is proposed. The proposed system is used for the anomaly detection of a motion sound-based actuator module. The proposed flow-based model, which is a generative model, learns by directly modeling a data distribution function. As the objective function is the maximum likelihood value of the input data, the training is stable and simple to use for anomaly detection. The operation sound of a car's side-view mirror motor is converted into a Mel-spectrogram image, consisting of a folding signal and an unfolding signal, and used as training data in this experiment. The proposed system is composed of an encoder and a decoder. The data extracted from the layer of the pretrained feature extractor are used as the decoder input data in the encoder. This information is used in the decoder by performing an interlayer cross-scale convolution operation. The experimental results indicate that the context information of various dimensions extracted from the interlayer hierarchical data improves the defect detection accuracy. This paper is notable because it uses acoustic data and a normalizing flow model to detect outliers based on the features of experimental data.