• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2015.10a
• /
• pp.903-905
• /
• 2015

The need for digital devices is increasing in the digital age. In general, noise in images occurs during the process of compression, recognition and processing due to many reasons. Some of the filters used to remove salt and pepper noise include SMF, CWMF and AMF. In areas where the noise density is high, the removal of noise is undermined. This paper suggests an adjusted median filter algorithm that preserves the non-noise pixels while transforming the noise pixels to more effectively remove salt and pepper noise.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2015.10a
• /
• pp.900-902
• /
• 2015

In the modern society, according to the advancement in digital image processing technology, edge detection is being utilized in various application sectors such as smart device and medical, etc. In existing edge detection methods, there are Sobel, Prewitt, Roberts and Laplacian, etc, which uses the mask. These previous methods are easy to implement but shows somewhat insufficient results. Therefore, in order to compensate the problems of existing methods, in this paper, an algorithm that detects the edge using the local mask and morphological operation was proposed and the detection performance was compared against the previous methods.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.21 no.6
• /
• pp.1167-1173
• /
• 2017

As the society increasingly embraces the high - tech digital information age, the field of image processing becomes progressively more branched out and becoming an imperative field. However, image data is deteriorated due to various causes during transmission and salt and pepper noise is typical. Typical methods for removing salt and pepper noise include CWMF, SWMF, and A-TMF. However, existing methods are somewhat insufficient in their ability to remove noise in salt and pepper noise environments. Therefore, in this paper, after it is determined whether noise removal is needed, the following measures were taken. If the center pixel was non-noise, the original pixel was preserved, If it was noise, we proposed a two - dimensional spline interpolation method and a median filter depending on the noise density of the local mask. For the purpose of objective judgment, we compared the results with that of existing methods and used PSNR (peak signal to noise ratio) as a judgment criterion.

• Journal of Internet Computing and Services
• /
• v.22 no.2
• /
• pp.59-68
• /
• 2021

As artificial intelligence(AI) technologies, which have made rapid growth recently, began to be applied to the marine environment such as ships, there have been active researches on the application of CNN-based models specialized for digital videos. In E-Navigation service, which is combined with various technologies to detect floating objects of clash risk to reduce human errors and prevent fires inside ships, real-time processing is of huge importance. More functions added, however, mean a need for high-performance processes, which raises prices and poses a cost burden on shipowners. This study thus set out to propose a method capable of processing information at a high rate while maintaining the accuracy by applying Quantization techniques of a deep learning model. First, videos were pre-processed fit for the detection of floating matters in the sea to ensure the efficient transmission of video data to the deep learning entry. Secondly, the quantization technique, one of lightweight techniques for a deep learning model, was applied to reduce the usage rate of memory and increase the processing speed. Finally, the proposed deep learning model to which video pre-processing and quantization were applied was applied to various embedded boards to measure its accuracy and processing speed and test its performance. The proposed method was able to reduce the usage of memory capacity four times and improve the processing speed about four to five times while maintaining the old accuracy of recognition.

• Journal of Broadcast Engineering
• /
• v.27 no.1
• /
• pp.3-12
• /
• 2022

Image deblurring aims to remove image blur, which can be generated while shooting the pictures by the movement of objects, camera shake, blurring of focus, and so forth. With the rise in popularity of smartphones, it is common to carry portable digital cameras daily, so image deblurring techniques have become more significant recently. Originally, image deblurring techniques have been studied using traditional optimization techniques. Then with the recent attention on deep learning, deblurring methods based on convolutional neural networks have been actively proposed. However, most of them have been developed while focusing on better performance. Therefore, it is not easy to use in real situations due to the speed of their algorithms. To tackle this problem, we propose a novel deep learning-based deblurring algorithm that can be operated in real-time on HD resolution. In addition, we improved the training and inference process and could increase the performance of our model without any significant effect on the speed and the speed without any significant effect on the performance. As a result, our algorithm achieves real-time performance by processing 33.74 frames per second at 1280×720 resolution. Furthermore, it shows excellent performance compared to its speed with a PSNR of 29.78 and SSIM of 0.9287 with the GoPro dataset.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.18 no.5
• /
• pp.1103-1109
• /
• 2014

The motion recognition system has been broadly studied in digital image and video processing fields. Recently, method using th depth image is used very useful. However, recognition accuracy of depth image based method will be loss caused by size and shape of object distorted for angle of the depth sensor. Therefore, distortion correction of depth sensor is positively necessary for distinguished performance of the recognition system. In this paper, we propose a pre-processing algorithm to improve the motion recognition system. Depth data from depth sensor converted to real world, performed the corrected angle, and then inverse converted to projective world. The proposed system make progress using the OpenCV and the window program, and we test a system using the Kinect in real time. In addition, designed using Verilog-HDL and verified through the Zynq-7000 FPGA Board of Xilinx.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.40 no.6
• /
• pp.51-60
• /
• 2003

The watermarking is the technique that embeds or extracts the certain data without the change of the original data for the copyright protection of the multimedia contents. Watermark-embedded contents must not be distinguished by human's eye and must be robust to the various image processing and the intentional distortions. In this paper, we propose a new watermarking technique applied in the wavelet domain which has both the spatial and frequency information of a image. For both the robustness and the invisibility, the positions for embedding the watermark is selected with the multi-threshold. We search the similarity between highly correlated coefficients in the each subband and decide the mark space after verifying the significance in the specified subband. The similarity is represented by the coefficient difference between the subbands and its distribution is used in the watermark embedding and extracting. The embedded watermark can be extracted without the original image using the relationship of the subbands. By these properties the proposed watermarking algorithm has the invisibility and the robustness to the attacks such as JPEG compression and the general image processing.

• Journal of Digital Contents Society
• /
• v.16 no.1
• /
• pp.157-164
• /
• 2015

The existing input device is limited to keyboard and mouse. However, recently new type of input device has been developed in response to requests from users. To reflect this trend we propose the new type of input device that gives instruction as analyzing the hand motion of image without special device. After binarizing the skin color area using Cam-Shift method and tracking, it recognizes the hand motion by inputting the finger areas and the angles from the palm center point, which are separated through labeling, into four cardinal directions and counting them. In cases when specific background was not set and without gloves, the recognition rate remained approximately at 75 percent. However, when specific background was set and the person wore red gloves, the recognition rate increased to 90.2 percent due to reduction in noise.

• Journal of Digital Policy
• /
• v.2 no.4
• /
• pp.33-39
• /
• 2023

With a continuous occurrence of right-turn traffic accidents at intersections, there is an increasing demand for measures to address these incidents. In response, a technology has been developed to detect the presence of pedestrians through object detection in CCTV footage at right-turn areas and display warning messages on the screen to alert drivers. The YOLO (You Only Look Once) model, a type of object detection model, was employed to assess the performance of object detection. An algorithm was also devised to address misidentification issues and generate warning messages when pedestrians are detected. The accuracy of recognizing pedestrians or objects and outputting warning messages was measured at approximately 82%, suggesting a potential contribution to preventing right-turn accidents

• Journal of Digital Convergence
• /
• v.17 no.10
• /
• pp.265-271
• /
• 2019

The number of galleries across the country is increasing as interest in cultural life increases due to the increase in national income. However, museum satisfaction is relatively low compared to other services. In this paper, we propose a service that provides preference information based on machine learning in embedded system environment in order to increase museum satisfaction. The proposed algorithm implements an embedded system using Raspberry Pi. Machine learning was used to find works similar to the viewer's favorite works, and several models were compared to select models applicable to embedded systems. By using the preference information, it is possible to effectively organize the gallery exhibition contents to increase the exhibition satisfaction and the re-visit rate of the museum.