• Journal of the Korea Institute of Military Science and Technology
• /
• v.26 no.3
• /
• pp.234-245
• /
• 2023

The Arena Fragmentation Test(AFT) is designed to analyze warhead performance by measuring fragmentation data. In order to evaluate the results of the AFT, a set of AFT images are captured by high-speed cameras. To detect objects in the AFT image set, ResNet-50 based Faster R-CNN is used as a detection model. However, because of the low resolution of the AFT image set, a detection model has shown low performance. To enhance the performance of the detection model, Super-resolution(SR) methods are used to increase the AFT image set resolution. To this end, The Bicubic method and three SR models: ZSSR, EDSR, and SwinIR are used. The use of SR images results in an increase in the performance of the detection model. While the increase in the number of pixels representing a fragment flame in the AFT images improves the Recall performance of the detection model, the number of pixels representing noise also increases, leading to a slight decreases in Precision performance. Consequently, the F1 score is increased by up to 9 %, demonstrating the effectiveness of SR in enhancing the performance of the detection model.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.17 no.7
• /
• pp.1794-1806
• /
• 2023

This study presents a method for capturing photographs of users as input and converting them into 2D character animation sprites using a generative adversarial network-based artificial intelligence network. Traditionally, 2D character animations have been created by manually creating an entire sequence of sprite images, which incurs high development costs. To address this issue, this study proposes a technique that combines motion videos and sample 2D images. In the 2D sprite generation process that uses the proposed technique, a sequence of images is extracted from real-life images captured by the user, and these are combined with character images from within the game. Our research aims to leverage cutting-edge deep learning-based image manipulation techniques, such as the GAN-based motion transfer network (impersonator) and background noise removal (U² -Net), to generate a sequence of animation sprites from a single image. The proposed technique enables the creation of diverse animations and motions just one image. By utilizing these advancements, we focus on enhancing productivity in the game and animation industry through improved efficiency and streamlined production processes. By employing state-of-the-art techniques, our research enables the generation of 2D sprite images with various motions, offering significant potential for boosting productivity and creativity in the industry.

• The Journal of Korea Institute of Information, Electronics, and Communication Technology
• /
• v.6 no.1
• /
• pp.6-17
• /
• 2013

Recently, the color of vehicle license plate has been changed from green to white. Thus the vehicle plate recognition system used for parking management systems, speed and signal violation detection systems should be robust to the both colors. This paper presents a vehicle license plate recognition system, which works on both of green and white plate at the same time. In the proposed system, the image of license plate is taken from a captured vehicle image by using morphological information. In the next, each character region in the license plate image is extracted based on the vertical and horizontal projection of plate image and the relative position of individual characters. Finally, for the recognition process of extracted characters, PCA(Principal Component Analysis) and LDA(Linear Discriminant Analysis) are sequentially utilized. In the experiment, vehicle license plates of both green background and white background captured under irregular illumination conditions have been tested, and the relatively high extraction and recognition rates are observed.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.18 no.5
• /
• pp.120-124
• /
• 2017

The temporal super resolution is a method for increasing the frame rate. Electro-optical sensors are used in various surveillance and reconnaissance weapons systems, and the spatial resolution and temporal resolution of the required electro-optical sensors vary according to the performance requirement of each weapon system. Because most image sensors capture images at 30~60 frames/second, it is necessary to increase the frame rate when the target moves and changes rapidly. This paper proposes a method to increase the frame rate using color channel extrapolation. Using a DMD, one frame of a general camera was adjusted to have different consecutive exposure times for each channel, and the captured image was converted to a single channel image with an increased frame rate. Using the optical flow method, a virtual channel image was generated for each channel, and a single channel image with an increased frame rate was converted to a color channel image. The performance of the proposed temporal super resolution method was confirmed by the simulation.

• Journal of Korean Society of Forest Science
• /
• v.109 no.1
• /
• pp.23-30
• /
• 2020

The use of increased computing power, machine learning, and deep learning techniques have dramatically increased in various sectors. In particular, image detection algorithms are broadly used in forestry and remote sensing areas to identify forest types and tree species. However, in South Korea, machine learning has rarely, if ever, been applied in forestry image detection, especially to classify tree species. This study integrates the application of machine learning and forest image detection; specifically, we compared the ability of two machine learning data collection methods, namely image data captured by forest experts (D1) and web-crawling (D2), to automate the classification of five trees species. In addition, two methods of characterization to train/test the system were investigated. The results indicated a significant difference in classification accuracy between D1 and D2: the classification accuracy of D1 was higher than that of D2. In order to increase the classification accuracy of D2, additional data filtering techniques were required to reduce the noise of uncensored image data.

• Journal of Broadcast Engineering
• /
• v.16 no.3
• /
• pp.561-565
• /
• 2011

This paper presents an imaging system for reconstruction of enhanced color images using the conventional Bayer CFA. By extracting various colors such as RGBCY from two sequential images which consist of a image by broadband G channel lens filter and the other image captured without one, the proposed color image reconstruction system can reduce the computational complexity for demosaicking and make high resolution color information without aliasing artifacts. Because the proposed system uses the common Bayer CFA image sensor, fabricating a new type of CFA is not necessary for obtaining a multi-spectral image, which can be easily extensible for applications of multi-spectral imaging. Finally, in order to verify the performance of the proposed system, experimental results are performed. By comparing with the existing demosaicking methods, the proposed camera system showed the significant improvements in the sense of color resolution.

• Journal of the Korean Society for Precision Engineering
• /
• v.27 no.2
• /
• pp.70-77
• /
• 2010

This study proposed an auto focusing method for a multi-focus image in assembling lens modules in digital camera phones. A camera module in a camera phone is composed of a lens barrel, an IR glass, a lens mount, a PCB board and aspheric lenses. Alignment among the components is one of the important factors in product quality. Auto-focus is essential to adjust image quality of an IR glass in a lens holder, but there are two focal points in the captured image due to thickness of IR glass. So, sharpness, probability and a scale factor are defined to find desired focus from a multi-focus image. The sharpness is defined as clarity of an image. Probability and a scale factors are calculated using pattern matching with a registered image. The presented algorithm was applied to a lens assembly machine which has 5 axes, two vacuum chucks and an inspection system. The desired focus can be determined on the local maximum of the sharpness, the probability and the scale factor in the experiment.

• Applied Biological Chemistry
• /
• v.41 no.2
• /
• pp.160-165
• /
• 1998

Current discriminating technique of rice variety is known to be not objective till this time because of depending on naked eye of well trained inspector. DNA finger print method based on genetic character of rice has been indicated inappropriate for on-site application, because the method need much labor and skilled expert. The purpose of this study was to develops the identification technique of polished rice varieties using CCD camera images. To minimize the noise of the captured image, thresholding and median filtering were carried out, and edge was extracted from the image data. Image data after pretreatment of normalize and FFT(fast fourier transform) were used for library model and feedforward backpropagation neural network model. Image processing technique using CCD camera could discriminate the variety of rice with high accuracy in case of quite different rice of shape, but the accuracy was reached at 85% in the similar shape of rice.

• Journal of Science of Art and Design
• /
• v.2
• /
• pp.193-221
• /
• 2000

The postmodern culture since the later 20C breaks downa tradition a relation between the reality and languages or sign images expressing it. It develops in the way to review the meaning on the object's imitation or the representation to have been followed since Plato and represent the new state and concept of expressed things. Also, The visual art leads an change of paradigm by images giving up the visual resemblance or the function of representation and endowing them with the new sense. This essay has a purpose to study an important discussion about this change centered on Baudrillard and Lacan. A sociologist Baudrillard promotes the concept of 'simulation' through detecting the reality and the social and historical state of the image. Studying on the course of this change, he calls the step that the image escapes from the stage to reflect the reality and become the pure imitation by itself simulation. The image in the stage of simulation is called 'hyperreality' because it doesn't have any an indicator or a substitute and happens by models without the original or the reality. So he asserts that art is not to contain some absoluteness or transcendency as the past, but to be as the spectacle with characteristics of meaningless, emptiness, contingency. Lacan dismantles the concept of the absolute Cogito to have become the center of the western ideology, and creates the concept of 'Other'. He concludes also the reality exists but can't be captured, and it's impossible for the thinking subject can reach it. The concept of new image which can be thought as the Symbolic in Lacan is 'Signifier without Signified' since it isn't possible to be the transcendent Signifier fixing the meaning finally in it. His 'Gaze' theory is which to be emitted in other's area determines the subject. Equally Baudrillard and Lacan sets up the new state of the image through the end of representation system As for Baudrillard, art intends to the worthlessness and is nothing but imagination. But in Lacan a picture represents the subject being in process by the dialectic of desire.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.42 no.5 s.305
• /
• pp.63-72
• /
• 2005

This paper proposes a new method to construct a panorama image from video sequences captured by the video camcoder revolving on the center axis of the tripod. The proposed method is consisted of two algorithms; frame selection and image mosaics. In order to select frames to construct the panorama image, we employ the interpolation search using the information in overlapped areas. This method can search suitable frames quickly. We construct an image mosaic using the projective transform induced from four pairs of quasi-features. The conventional methods select feature points by using only texture information, but the presented method in this paper uses the position of each feature point as well. We make an experiment on the proposed method with real video sequences. The results show that the proposed method is better than the conventional one in terms of image quality.