• Journal of IKEEE
• /
• v.25 no.3
• /
• pp.493-500
• /
• 2021

In this paper, we propose the development of deep learning structure to improve quality of polygonal containers. The deep learning structure consists of a convolution layer, a bottleneck layer, a fully connect layer, and a softmax layer. The convolution layer is a layer that obtains a feature image by performing a convolution 3x3 operation on the input image or the feature image of the previous layer with several feature filters. The bottleneck layer selects only the optimal features among the features on the feature image extracted through the convolution layer, reduces the channel to a convolution 1x1 ReLU, and performs a convolution 3x3 ReLU. The global average pooling operation performed after going through the bottleneck layer reduces the size of the feature image by selecting only the optimal features among the features of the feature image extracted through the convolution layer. The fully connect layer outputs the output data through 6 fully connect layers. The softmax layer multiplies and multiplies the value between the value of the input layer node and the target node to be calculated, and converts it into a value between 0 and 1 through an activation function. After the learning is completed, the recognition process classifies non-circular glass bottles by performing image acquisition using a camera, measuring position detection, and non-circular glass bottle classification using deep learning as in the learning process. In order to evaluate the performance of the deep learning structure to improve quality of polygonal containers, as a result of an experiment at an authorized testing institute, it was calculated to be at the same level as the world's highest level with 99% good/defective discrimination accuracy. Inspection time averaged 1.7 seconds, which was calculated within the operating time standards of production processes using non-circular machine vision systems. Therefore, the effectiveness of the performance of the deep learning structure to improve quality of polygonal containers proposed in this paper was proven.

• Journal of Korean Society of Forest Science
• /
• v.101 no.2
• /
• pp.266-278
• /
• 2012

The objective of this study was developing dynamic stand growth model to predict diameter at breast height (DBH) growth by thinning intensity and cycle for major tree species of South Korea. The yield table, one of static stand growth models, constructed by Korea Forest Service was employed to prepare dynamic stand growth models for 8 tree species. In the process of model development, the thinning type was designated to thinning from below and equations for predicting the DBH change after thinning by different intensities was generated. In addition, stand density (N/ha), age and site index were adopted as explanatory variables for DBH prediction model. Thereafter, using the model, DBH growth under various silvicuture through integrating such equations considering thinning intensities, and cycles. The dynamic stand growth model of DBH developed in this study can provide understanding of effectiveness in forest growth and growing stock when thinning practice is performed in forest. Furthermore, results of this study is also applicable to quantitatively assess the carbon storage sequestration capability.

• Journal of the Korean Society of Radiology
• /
• v.14 no.4
• /
• pp.487-494
• /
• 2020

The distance between the source and the detector, the diameter of the detector, and the volume effect of the radiation source result in a change in solid angle at the detector entrance, which affects the determination of detection efficiency by causing a difference in path length within the detector. A typical analysis method for calculating solid angles was useful only for a source (⁶⁰Co) with a simple geometric structure, so in this experiment, the distance between the detector and the source was measured by switching on for up to 25 cm with the reference point of window cap 0.5 cm. In addition, 450 and 1000 ㎖ Marinelli beaker of standard volumetric sources were closely adhered to the detector. For circular point sources co-axial with the detector, the change in the solid angle to the distance from the detector window is equal to half the square radius of the source versus the square radius of the detector, if the resulting relationship of the calculation analysis results in the detector being less than the radius of the source. Since the solid angular difference is 0.5 the result of Monte Carlo is acceptable. The relationship between detector and source distance is shown. Solid angles have been verified to decrease rapidly with distance. Measurement and simulation results for a volumetric source show a difference of ±1.01% from a distance of 0 cm and less than 4 % when the distance is reduced to 5 and 10 cm. It can be seen that the longer distance, the smaller efficiency angle, and the exponential increase in attenuation as the energy decreases, is reflected in the calculation of efficiency. Thus, the detection efficiency has proved sufficient for the use of solid angle and Monte Carlo codes.