• Journal of the Korea Society of Computer and Information
• /
• v.26 no.7
• /
• pp.19-27
• /
• 2021

In this paper, we propose a method for detecting characters in images and detecting facial regions, which consists of two tasks. First, we separate two different characters to detect the face position of the characters in the frame. For fast detection, we use You Only Look Once (YOLO), which finds faces in the image in real time, to extract the location of the face and mark them as object detection boxes. Second, we present three image processing methods to detect accurate face area based on object detection boxes. Each method uses HSV values extracted from the region estimated by the detection figure to detect the face region of the characters, and changes the size and shape of the detection figure to compare the accuracy of each method. Each face detection method is compared and analyzed with comparative data and image processing data for reliability verification. As a result, we achieved the highest accuracy of 87% when using the split rectangular method among circular, rectangular, and split rectangular methods.

• The Journal of the Korea Contents Association
• /
• v.16 no.8
• /
• pp.395-403
• /
• 2016

Recently, Image processing has been used in many areas. In the image processing techniques that a lot of research is tracking of moving object in real time. There are a number of popular methods for tracking an object such as HOG(Histogram of Oriented Gradients) to track pedestrians, and Codebook to subtract background. However, object extraction has difficulty because that a moving object has dynamic background in the image, and occurs severe lighting changes. In this paper, we propose a method of object extraction using depth image and color image features based on ROI(Region of Interest). First of all, we look for the feature points using the color image after setting the ROI a range to find the location of object in depth image. And we are extracting an object by creating a new contour using the convex hull point of object and the feature points. Finally, we compare the proposed method with the existing methods to find out how accurate extracting the object is.

• Journal of Korean Clinical Nursing Research
• /
• v.14 no.3
• /
• pp.5-14
• /
• 2008

Purpose: The purpose of this study was to compare body image, depression, and climacteric symptoms among middle-aged women with and without thyroidectomy. Method: The research design was a comparative descriptive study using a self-report questionnaire. A total of 102 women aged 40~59 years were recruited using a convenient sampling method. A total of 52 women out of 102 were within a year after thyroidectomy due to benign disease. The rest of the subjects were healthy women without any previous diseases in thyroid. The instruments included Body Image Concern Inventory, Climacteric symptoms, and Center for Epidemiological Studies Depression Scale. The data were analyzed by descriptive statistics, t-test, $x^2$-test, one-way ANOVA, and correlation analysis. Results: There were no significantly differences in age, education, menstruation states between women with and without thyroidectomy. The demographic characteristics of two groups were homogeneous. There were significantly statistical differences in body image (t= 8.456, p=.000), and depression (t=3.142, p=.002) between the groups. There was, however, no significantly differences of climacteric symptoms. In addition, in women with thyroidectomy, no significant associations were found among body image, depression, and climacteric symptoms. Conclusion: The findings demonstrated the need to develop an effective nursing intervention to increase body image and to reduce depression in middle-aged women after thyroidectomy.

• Journal of the Institute of Convergence Signal Processing
• /
• v.1 no.1
• /
• pp.82-88
• /
• 2000

In this paper, We compare the Harr wavelet of the S+P transform with various integer coefficient filter banks and apply 9/7 ICFB to the wavelet transform. In addition, we propose a entropy-coding method that exploits the multiresolution structure and the feedback of the prediction error, and can efficiently compress the transformed image for progressive transmission. Simulation results are included to compare to the compression ratio using the S+P transform with different types of images.

• International Journal of Highway Engineering
• /
• v.18 no.4
• /
• pp.47-53
• /
• 2016

PURPOSES : The objective of this study was to evaluate the results of the dynamic immersion test (DIT) through digital image analysis (DIA). METHODS : The asphalt binder retained post DIT was inspected visually by more than three investigators. However, because visual observations can be subjective, depending on the inspector's skills and knowledge, DIA was also performed. The threshold value for the DIA was determined by a mesh analysis, in which the digital image to be analyzed is divided into very small meshes. In addition, the bitumen bond strength (BBS) test was also performed to the compare the visual results with the mechanical values. RESULTS AND CONCLUSIONS : Depending on the materials used, various methods can be used to predict the behavior of retained asphalt. However, the increasing ratio in the trend of retained asphalt shows different behaviors among the evaluation methods. In this study, the results of the visual observations were significantly different from those of the DIA, the mesh analysis, and the BBS tests. Thus, DIA is an appropriate method for evaluating the results of the DIT. However, in order to use this technique in the field, it is necessary to determine a more reasonable threshold value by performing DIA on various materials.

• Journal of Broadcast Engineering
• /
• v.23 no.3
• /
• pp.403-408
• /
• 2018

Generally, a large data set is required for learning of deep learning. However, since it is not easy to create large data sets, there are a lot of techniques that make small data sets larger through data expansion such as rotation, flipping, and filtering. However, these simple techniques have limitation on extendibility because they are difficult to escape from the features already possessed. In order to solve this problem, we propose a method to acquire new image data by using existing data. This is done by retrieving and acquiring similar images using existing image data as a query of the content-based image retrieval (CBIR). Finally, we compare the performance of the base model with the model using CBIR.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.6 no.6
• /
• pp.909-914
• /
• 2002

In this thesis, noise reduction of image with impulse noise in circle image removed noise to harness existing median filter for noise reduction from image data of damage by noise when impulse noise is high or noise reduction is low, but it is not made up of noise reduction to harness existing median filter in case of existence of non-impulse noise. Therefore noise reduction of image with non-impulse noise had to remove noise by morphological arithmetic in this thesis's proposition. In contrast to median filtering, result of edge detection is more efficient after remove non-impulse noise by method of thesis's proposition and it compare and demonstrate through this experimentation.

• Journal of the Korea Society of Computer and Information
• /
• v.7 no.1
• /
• pp.112-119
• /
• 2002

We propose an enhanced wavelet packet image coder algorithm which is based on the coefficients partition. The proposed wavelet packet image coder uses the first-order entropy to reduce the total compression time, and achieves low bit rates and rate-distortion performance by the zero-tree based coding using correlations between coefficients partition. This new algorithm represents new parent-children relationships for reducing image reconstruction error using the correlations between each frequency subbands and then the wavelet packet coefficients are Partitioned by a new order. The computer simulations demonstrate higher PSNR under the same bit rate and improved image compression time and enhanced rate control compare with conventional algorithms. From the simulation results, it is shown that the encoding and decoding process of proposed coder are much simple and accurate than present method against texture images , which include many mid-frequency elements.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2016.10a
• /
• pp.953-955
• /
• 2016

Degradation is occurred in the process of the signal transmission in the image processing system due to various reasons. Degradation is noise addition in the image signal and the representative one to cause degradation is salt and pepper noise. Therefore, image restoring filter was suggested in this article to apply and process weighted value by the changes of each directional pixel upon breakdown of local mask with 8 directions in order to restore the damaged image in the environment of salt and pepper noise. In addition, peak signal to noise ratio (PSNR) was used to compare the existing method as the objective determinant standard of the improvement effect.

• Smart Media Journal
• /
• v.8 no.2
• /
• pp.46-57
• /
• 2019

In recent research, deep learning-based methods have achieved state-of-the-art performance in various computer vision tasks. However, these methods are commonly supervised, and require huge amounts of annotated data to train. Acquisition of data demands an additional costly effort, particularly for the tasks where it becomes challenging to obtain large amounts of data considering the time constraints and the requirement of professional human diligence. In this paper, we present a data level synthetic sampling solution to learn from small and imbalanced data sets using Generative Adversarial Networks (GANs). The reason for using GANs are the challenges posed in various fields to manage with the small datasets and fluctuating amounts of samples per class. As a result, we present an approach that can improve learning with respect to data distributions, reducing the partiality introduced by class imbalance and hence shifting the classification decision boundary towards more accurate results. Our novel method is demonstrated on a small dataset of 2789 tomato plant disease images, highly corrupted with class imbalance in 9 disease categories. Moreover, we evaluate our results in terms of different metrics and compare the quality of these results for distinct classes.