• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제15권12호
• /
• pp.4420-4438
• /
• 2021

Image dehazing is an ill-posed problem which is far from being solved. Traditional image dehazing methods often yield mediocre effects and possess substandard processing speed, while modern deep learning methods perform best only in certain datasets. The haze removal effect when processed by said methods is unsatisfactory, meaning the generalization performance fails to meet the requirements. Concurrently, due to the limited processing speed, most dehazing algorithms cannot be employed in the industry. To alleviate said problems, a lightweight fast dehazing network based on a multiple scale-patch framework (MSP) is proposed in the present paper. Firstly, the multi-scale structure is employed as the backbone network and the multi-patch structure as the supplementary network. Dehazing through a single network causes problems, such as loss of object details and color in some image areas, the multi-patch structure was employed for MSP as an information supplement. In the algorithm image processing module, the image is segmented up and down for processed separately. Secondly, MSP generates a clear dehazing effect and significant robustness when targeting real-world homogeneous and nonhomogeneous hazy maps and different datasets. Compared with existing dehazing methods, MSP demonstrated a fast inference speed and the feasibility of real-time processing. The overall size and model parameters of the entire dehazing model are 20.75M and 6.8M, and the processing time for the single image is 0.026s. Experiments on NTIRE 2018 and NTIRE 2020 demonstrate that MSP can achieve superior performance among the state-of-the-art methods, such as PSNR, SSIM, LPIPS, and individual subjective evaluation.

• Journal of the Korea Convergence Society
• /
• 제8권12호
• /
• pp.85-92
• /
• 2017

As accessibility toward traditional cultural contents drops compared to its increase in production, the need for higher accessibility for continued management and research to exist. For this, this paper introduces an image classifier model for traditional images based on artificial neural networks, which converts the input image's features into a vector space and by utilizing a RNN based model it recognizes and compares the details of the input which enables the classification of traditional images. This enables the classifiers to classify similarly looking traditional images more precisely by focusing on the details. For the training of this model, a wide range of images were arranged and collected based on the format of the Korean information culture field, which contributes to other researches related to the fields of using traditional cultural images. Also, this research contributes to the further activation of demand, supply, and researches related to traditional culture.

• Archives of design research
• /
• 제16권4호
• /
• pp.481-488
• /
• 2003

A method to select the proper file formats of thumbnail images is proposed. After the experimental works for image file formats such as JPEG, GIF, and those effectiveness to the features contained in images, four features are obtained by feature extraction methods used in contents based image indexing, those are, the details, highly saturated colored area, the number of clustered color, and the amount of continuously varying hue. Also it is described the way to select the proper file format with those four features. In the thumbnail image generation experiments, 6 sample images are used, and with subjective assessment experiments, the resulted thumbnail images are shown to be consistent to the file formats chosen by human subjects, that is, favorable to human vision, which means the proposed method can be utilized as an automatic and systematic generation of thumbnail images for a lot of images on Web.

• Korean Institute of Interior Design Journal
• /
• 제19권6호
• /
• pp.179-187
• /
• 2010

For the user-centered interior space design, a designer needs to carefully focus on the details from which the user must keenly observe or concentrate on the space and how to deliberately check the image of the space. Following this view, for the user-centered interior space design this study analyzes the way which the space is represented from the perspective of the image assessment. The results obtained from this study are summarized as follows: First, this study analyzed the characteristic of space image change in the form of cross-comparison between one space and a rotated space. With analyzing of an eye fixation by showing the space, the space images of "concentration-dispersion" and "strengthening-weaken "have an important role in the analysis of the perception of space. it confirmed that the method of space perception was changed by rotated the space. Second, With changing quantities of image and extraction of deviation from adjective in survey, it quantitatively graps that respondents of feel in space perception by changing the space and "concentration" and 'dispersion" for process of space choice. The results of the research can provide important basis to judge the changing of space perception by visual perception. Third, through the analysis of image change rate and deviation rate, the characteristics of image change with space change can be analyzed. The results derived from the study provide the evidence to support the image change by space rotating.

• Journal of Biosystems Engineering
• /
• 제26권6호
• /
• pp.545-552
• /
• 2001

Nondestructive methods such as ultrasonic and magnetic resonance imaging systems have many advantages but still much expensive. And they do not give exact color information and may miss some details. If it is allowed to destruct a biological object to obtain interior and exterior informations, 3D image visualization model from a series of sliced sectional images gives more useful information with relatively low cost. In this paper, a PC based automatic 3D visualization system is presented. The system is composed of three modules. The first module is the handling and image acquisition module. The handling module feeds and slices a cylindrical shape paraffin, which holds a biological object inside the paraffin. And the paraffin is kept being solid by cooling while being handled. The image acquisition modulo captures the sectional image of the object merged into the paraffin consecutively. The second one is the system control and interface module, which controls actuators for feeding, slicing, and image capturing. And the last one is the image processing and visualization module, which processes a series of acquired sectional images and generates a 3D volumetric model. To verify the condition for the uniform slicing, normal directional forces of the cutting edge according to the various cutting angles were measured using a strain gauge and the amount of the sliced chips were weighed and analyzed. Once the 3D model was constructed on the computer, user could manipulate it with various transformation methods such as translation, rotation, and scaling including arbitrary sectional view.

• Journal of Broadcast Engineering
• /
• 제19권1호
• /
• pp.96-108
• /
• 2014

In case of contrast of the image enhancement by using the conventional histogram equalization, over-enhancement, false contouring and distortion such as the details disappearance of the image occurs due to the excessive brightness change. Especially, these distortion appears when the brightness distribution is concentrated in a particular brightness level. In order to solve these problems, improved histogram equalization methods to transform the input histogram by clipping using threshold have been proposed, but contrast enhancement effect is reduced because it does not consider the characteristics of the input image's histogram to apply the same threshold for the entire histogram, and unnatural image is obtained because it does not retain the characteristics of the image. In this paper, to solve the problems of existing methods, we propose new equalization method that suppress excessive brightness changes by applying to the differential compression according to the histogram frequency, and maintain the characteristics of the input image. In addition, we propose a more effectively method to improve contrast by controlling the strength of the compression ratio depending on the characteristics of the input image.

• Journal of IKEEE
• /
• 제23권3호
• /
• pp.910-916
• /
• 2019

Non-local similarity of natural images is one of highly exploited features in various applications dealing with images. Unique edges, texture, and pattern of the images are frequently repeated over the entire image. Once the similar image blocks are classified into a cluster, representative features of the image blocks can be extracted from the cluster. The bigger the size of the cluster is the better the additive white noise can be separated. Denoising is one of major research topics in the image processing field suppressing the additive noise. In this paper, a denoising algorithm is proposed which first clusters the noisy image blocks based on similarity, extracts the feature of the cluster, and finally recovers the original image. Performance experiments with several images under various noise strengths show that the proposed algorithm recovers the details of the image such as edges, texture, and patterns while outperforming the previous methods in terms of PSNR in removing the additive Gaussian noise.

• Journal of the Korean Society for Precision Engineering
• /
• 제24권9호
• /
• pp.140-147
• /
• 2007

As the use of robots in surgeries becomes more frequent, the registration of medical devices based on images becomes more important. This paper presents two numerical algorithms for the registration of cross-sectional medical images such as CT (Computerized Tomography) or MRI (Magnetic Resonance Imaging) by using the geometrical information from helix or line fiducials. Both registration algorithms are designed to be used for a surgical robot that works inside a cavity of human body. This paper also reports details about the fiducial pattern that includes four helices and one line. The algorithms and the fiducial pattern were tested in various computer-simulated situations, and the results showed excellent overall registration accuracy.

• Proceedings of the Korea Institute of Convergence Signal Processing
• /
• 한국신호처리시스템학회 2005년도 추계학술대회 논문집
• /
• pp.167-170
• /
• 2005

Recently, filtering methods for attenuating noise while preserving image details are in progress actively. And SM(standard median) filter showed a great performance for noise removal in impulse noise environment but, it caused edge cancellation error. So, variable methods that modified SM(standard median) filter have been proposed, and CWM(center weighted median) filter is representative. Also, there are several methods to improve the efficiency based on min/max operation in term of preserving detail and filtering speed. In this paper, we managed a pixel corrupted by impulsive noise using min/max value of the surrounding band enclosing a pixel, and compared the efficiency with exiting methods in the simulation.

• Journal of the Korean Institute of Telematics and Electronics S
• /
• 제34S권7호
• /
• pp.89-101
• /
• 1997

A technical method of noise modeling and adaptive filtering reducing of speckle noise in ultrasonic medical images is presented. By adjusting the characteristics of the filer according to local statistics around each pixel of the image as moving windowing, it is possible to suppress noise sufficiently while preserve edge and other significant information required in diagnosis. Homogeneous factor(HF) from the noise models that enables the filter to recognize the local structures of the image is introduced, and an algorithm for determining the HF fitted to the diagnostic systems with various inner statistical properties is proposed. We show by the experimented that the performance of proposed method is superior to these of other filters and models in preserving small details and suppressing the noise at homogeneous region.