• Korean Journal of Remote Sensing
• /
• v.23 no.1
• /
• pp.7-14
• /
• 2007

Atmospheric correction is one of critical procedures to extract quantitative information related to biophysical variables from hyperspectral imagery. Most atmospheric correction algorithms developed for hyperspectral data have been based upon atmospheric radiative transfer (RT) codes, such as MODTRAN. Because of the difficulty in acquisition of atmospheric data at the time of image capture, the complexity of RT model, and large volume of hyperspectral data, atmospheric correction can be very difficult and time-consuming processing. In this study, we attempted to develop an efficient method for the atmospheric correction of EO-1 Hyperion data. This method uses the pre-calculated look-up-table (LUT) for fast and simple processing. The pre-calculated LUT was generated by successive running of MODTRAN model with several input parameters related to solar and sensor geometry, radiometric specification of sensor, and atmospheric condition. Atmospheric water vapour contents image was generated directly from a few absorption bands of Hyperion data themselves and used one of input parameters. This new atmospheric correction method was tested on the Hyperion data acquired on June 3, 2001 over Seoul area. Reflectance spectra of several known targets corresponded with the typical pattern of spectral reflectance on the atmospherically corrected Hyperion image, although further improvement to reduce sensor noise is necessary.

• Journal of the Korea Society of Computer and Information
• /
• v.26 no.2
• /
• pp.19-25
• /
• 2021

In this paper proposes a method of finding missing persons based on face-recognition technology and deep learning. In this paper, a real-time face-recognition technology was developed, which performs face verification and improves the accuracy of face identification through data fortification for face recognition and convolutional neural network(CNN)-based image learning after the pre-processing of images transmitted from a mobile device. In identifying a missing person's image using the system implemented in this paper, the model that learned both original and blur-processed data performed the best. Further, a model using the pre-learned Noisy Student outperformed the one not using the same, but it has had a limitation of producing high levels of deflection and dispersion.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.44 no.2 s.314
• /
• pp.37-45
• /
• 2007

We propose an adaptive image enhancement method for epigraphic images, which is based on local statistics. Local statistics of the image are utilized for adaptive realization of the enhancement, that controls the contribution of the smoothing or sharpening paths. Image contrast enhancement occurs in details and noises are suppressed in smooth areas. For modeling the epigraphic image, pre~process is achieved by HSDI(Hanzi squeezed digital image). We have calculated the local statistics from this HSDI model. Application of this approach to HSDI has shown that processing not only smooths the background areas but also improves the subtle variations of edges, so that the word regions can be enhanced. Experimental results show that the proposed algorithm has better performance than the conventional image enhancement ones.

• Korean Journal of Remote Sensing
• /
• v.23 no.3
• /
• pp.171-179
• /
• 2007

Hyperspectral images have shown a great potential for the applications in resource management, agriculture, mineral exploration and environmental monitoring. However, due to the large volume of data, processing of hyperspectral images faces some difficulties. This paper introduces the development of an image processing tool (HYVIEW) that is particularly designed for handling hyperspectral image data. Current version of HYVIEW is dealing with efficient algorithms for displaying hyperspectral images, selecting bands to create color composites, and atmospheric correction. Three band-selection schemes for producing color composites are available based on three most popular indexes of OIF, SI and CI. HYVIEW can effectively demonstrate the differences in the results of the three schemes. For the atmospheric correction, HYVIEW utilizes a pre-calculated LUT by which the complex process of correcting atmospheric effects can be performed fast and efficiently.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2003.10a
• /
• pp.416-419
• /
• 2003

The cellular neural networks have the circuit structure that differs from the form of general neural network. It consists of an array of the same cell which is a simple processing element, and each of the cells has local connectivity and space invariant template property. In this paper, time-multiplex image processing technique is applied for processing large images using small size CNN cell block, and we simulate the edge detection of a large image using the simulator implemented with a c program and matlab model. A 5$\times$5 CNN hardware and pre post processor is also implemented and is under test.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2023.05a
• /
• pp.687-688
• /
• 2023

Bone age assessment is a crucial task in pediatric radiology for assessing growth and development in children. In this paper, we explore the potential of Vision Transformer, a state-of-the-art deep learning model, for bone age assessment using X-ray images. We generate heatmap outputs using a pre-trained Vision Transformer model on a publicly available dataset of hand X-ray images and show that the model tends to focus on the overall hand and only the bone part of the image, indicating its potential for accurately identifying the regions of interest for bone age assessment without the need for pre-processing to remove background noise. We also suggest two methods for extracting the region of interest from the heatmap output. Our study suggests that Vision Transformer holds great potential for bone age assessment using X-ray images, as it can provide accurate and interpretable output that may assist radiologists in identifying potential abnormalities or areas of interest in the X-ray image.

• International Journal of Computer Science & Network Security
• /
• v.21 no.11
• /
• pp.17-22
• /
• 2021

Recently, image super-resolution techniques used in convolutional neural networks (CNN) have led to remarkable performance in the research area of digital image processing applications and computer vision tasks. Convolutional layers stacked on top of each other can design a more complex network architecture, but they also use more memory in terms of the number of parameters and introduce the vanishing gradient problem during training. Furthermore, earlier approaches of single image super-resolution used interpolation technique as a pre-processing stage to upscale the low-resolution image into HR image. The design of these approaches is simple, but not effective and insert the newer unwanted pixels (noises) in the reconstructed HR image. In this paper, authors are propose a novel single image super-resolution architecture based on synchronized depthwise separable convolution with Dense Skip Connection Block (DSCB). In addition, unlike existing SR methods that only rely on single path, but our proposed method used the synchronizes path for generating the SISR image. Extensive quantitative and qualitative experiments show that our method (SDCN) achieves promising improvements than other state-of-the-art methods.

• Journal of the Korean Society of Manufacturing Technology Engineers
• /
• v.22 no.3_1spc
• /
• pp.496-503
• /
• 2013

Recently, the death rate owing to breast cancers has been increasing, and the occurrence age for breast cancers is lowering every year. Mammography is known to be a reliable detection method for breast cancers and works by detecting texture changes, calcifications, and other potential symptoms. In this research on breast cancer detection, candidate objects were detected by using image processing on mammograms, and feature analysis was used to classify candidate objects as benign tumors and malignant tumors. To find candidate objects, image pre-processing and binarization using multiple thresholds, and the grouping of micro-calcifications were used. More than 50 shape features and intensity features were used in the classification. The performance of the detection algorithm by using Euclidian distance method for benign tumors was 93%, and the classification error rate was approximately 2%.

• Journal of the Korean Wood Science and Technology
• /
• v.38 no.6
• /
• pp.498-506
• /
• 2010

To use domestic softwood for structural lumber, appropriate grading system for quality, production and distribution condition of domestic lumber should be prepared. Kim et al. developed an automatic image processing system for grading domestic structural lumber (2009a and b). This study was carried out to investigate optimal image merging conditions for improving performance of image input system which is the key technique of image processing system, developed in the previous paper. To merge digital images of Korean larch lumber, choosing the green channel information of obtained image data showed the most accurate merging performance. As a pre-treatment process, applying Y-derivative Sharr's kernel filter could improve the image merging accuracy, but the effect of camera calibration was imperceptible. The optimal size of template image was verified as 30 pixel widths and 150 pixel heights. When applying the above mentioned conditions, the error length of images was 3.1 mm and the processing time was 9.7 seconds in average.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.15 no.5
• /
• pp.1778-1797
• /
• 2021

Document image binarization is an important pre-processing step in document analysis and archiving. The state-of-the-art models for document image binarization are variants of encoder-decoder architectures, such as FCN (fully convolutional network) and U-Net. Despite their success, they still suffer from three limitations: (1) reduced feature map resolution due to consecutive strided pooling or convolutions, (2) multiple scales of target objects, and (3) reduced localization accuracy due to the built-in invariance of deep convolutional neural networks (DCNNs). To overcome these three challenges, we propose an improved semantic segmentation model, referred to as DP-LinkNet, which adopts the D-LinkNet architecture as its backbone, with the proposed hybrid dilated convolution (HDC) and spatial pyramid pooling (SPP) modules between the encoder and the decoder. Extensive experiments are conducted on recent document image binarization competition (DIBCO) and handwritten document image binarization competition (H-DIBCO) benchmark datasets. Results show that our proposed DP-LinkNet outperforms other state-of-the-art techniques by a large margin. Our implementation and the pre-trained models are available at https://github.com/beargolden/DP-LinkNet.