• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.17 no.3
• /
• pp.740-753
• /
• 2023

Text recognition in natural scene images is a challenging problem in computer vision. The accurate identification of ship number characters can effectively improve the level of ship traffic management. However, due to the blurring caused by motion and text occlusion, the accuracy of ship number recognition is difficult to meet the actual requirements. To solve these problems, this paper proposes a dual-branch network based on the CRNN identification network. The network couples image restoration and character recognition. The CycleGAN module is used for blur restoration branch, and the Pix2pix module is used for character occlusion branch. The two are coupled to reduce the impact of image blur and occlusion. Input the recovered image into the text recognition branch to improve the recognition accuracy. After a lot of experiments, the model is robust and easy to train. Experiments on CTW datasets and real ship maps illustrate that our method can get more accurate results.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2021.10a
• /
• pp.121-123
• /
• 2021

Image processing is being used in various forms in important fields of the 4th industrial revolution, such as artificial intelligence, smart factories, and the IoT industry. In particular, in systems that require data processing such as object tracking, medical images, and object recognition, noise removal is used as a preprocessing step, but the existing algorithm has a drawback in that blurring occurs in the filtering process. Therefore, in this paper, we propose a filter algorithm using switching fuzzy weights. The proposed algorithm switches the fuzzy function by dividing the low-frequency region and the high-frequency region by the standard deviation of the filtering mask, and obtains the final output according to the fuzzy weight. The proposed algorithm showed improved results compared to the existing method, and showed excellent characteristics in the region where the high-frequency component is strong.

• Nuclear Engineering and Technology
• /
• v.55 no.12
• /
• pp.4591-4596
• /
• 2023

Single-photon emission computed tomography SPECT image reconstruction methods have a significant influence on image quality, with filtered back projection (FBP) and ordered subset expectation maximization (OSEM) being the most commonly used methods. In this study, we proposed newly-designed adaptive non-blind deconvolution with a structural similarity (SSIM) index that can take advantage of the FBP and OSEM image reconstruction methods. After acquiring brain SPECT images, the proposed image was obtained using an algorithm that applied the SSIM metric, defined by predicting the distribution and amount of blurring. As a result of the contrast to noise ratio (CNR) and coefficient of variation evaluation (COV), the resulting image of the proposed algorithm showed a similar trend in spatial resolution to that of FBP, while obtaining values similar to those of OSEM. In addition, we confirmed that the CNR and COV values of the proposed algorithm improved by approximately 1.69 and 1.59 times, respectively, compared with those of the algorithm involving an inappropriate deblurring process. To summarize, we proposed a new type of algorithm that combines the advantages of SPECT image reconstruction techniques and is expected to be applicable in various fields.

• Journal of the Korean Society of Radiology
• /
• v.14 no.6
• /
• pp.763-772
• /
• 2020

During X-ray examinations in the DR system, the scattered X-rays physically generated by the patient cause image blurring in poor quality. Although X-rays to increase the contrast of images, this increases the patient's exposure dose and is likely to result in grid induced artifacts. Therefore, the purpose of this study is obtain images similar to those of real-grid with non-grid level conditions using a VGR (Virtual Grid Role) algorithm that serves as a virtual grid. Comparing MTF, SNR and CNR of non-grid and VGR algorithm images obtained with 70% exposure conditions of real-grid images showed that the MTF0.5 differed from 0.265 to 0.350 and the MTF0.1 from 0.412 to 0.467 and the SNR, CNR were also different. In addition, comparing MTF, SNR and CNR of VGR algorithm and real-grid images showed that the MTF0.5 differed from 0.350 to 0.367 and the MTF0.1 from 0.467 to 0.483 and the SNR, CNR by little.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.14 no.10
• /
• pp.2261-2268
• /
• 2010

The information captured by imaging devices such as CCD or CIS may contain external noises through the processes of passing signals or storing images. In this paper, we propose a Modified Sigma Filter (MSF) algorithm to reduce such noises. In experiment, we verified that our MSF algorithm showed better performance in PSNR and 1D plot of simulation results compared with Gaussian Filter (GF), Local Sigma Filter (LSF). Tested images include random Gaussian Noises.

• Journal of Satellite, Information and Communications
• /
• v.7 no.3
• /
• pp.145-148
• /
• 2012

In applying LCD to TV application, one of the most significant factors to be improved is image sticking on the moving picture. LCD is different from CRT in the sense that it's continuous passive device, which holds images in entire frame period, while impulse type device generate image in very short time. To reduce image sticking problem related to hold typedisplay mode, we made an experiment to drive TN-LCD like CRT. We made articulate images by turn on-off backlight, and we realized the ratio of Back Light on-off time by counting between on time and off time for video signal input during 1 frame (16.7ms). Conventional CCFL (cold cathode fluorescent lamp) cannot follow fast on-off speed, so we evaluated new fluorescent substances of light source to improve residual light characteristic of CCFL. We realized articulate image generation similar to CRT by CCFL blinking drive and TN-LCD overdriving. As a result, reduced image sticking phenomenon was validated by naked eye and response time measurement.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.37 no.2A
• /
• pp.105-112
• /
• 2012

Face detection at a distance faces is very challenging since images are often degraded by blurring and noise as well as low resolution. This paper proposes an improved face detection method with AdaBoost filtering and sequential testing stages with color and shape information. The conventional AdaBoost filter detects face regions but often generates false alarms. The face detection method is improved by adopting sequential testing stages in order to remove false alarms. The testing stages comprise skin-color test and variable edge-mask filtering. The skin-color filtering is composed of two steps, which involve rectangular window regions and individual pixels to generate binary face clusters. The size of the variable edge-mask is determined by the ellipse which is estimated from the face cluster. The validation of the horizontal and vertical ratio of the mask is also investigated. In the experiments, the efficacy of the proposed algorithm is proved by images captured by a CCTV and a smart-phone

• Journal of the Korea Society of Computer and Information
• /
• v.16 no.12
• /
• pp.43-48
• /
• 2011

Among various methods proposed earlier, fuzzy image filtering is usually one of the favored techniques because it has less blurring effect and the decrease of noise removal rate after filtering. However, fuzzy filtering is ineffective on color images since it is firstly developed with gray scale. Thus, in this paper, we propose a fuzzy filtering algorithm for color images. First, we divide RGB color information from image into three channels of R, G, and B and judge the possibility of each pixel with mask by fuzzy logic independently. The output pixel value might be the average or median according to the degree of noise. Our experiment successfully verifies the effectiveness of new algorithm in color image.

• Journal of Korea Multimedia Society
• /
• v.5 no.2
• /
• pp.167-175
• /
• 2002

Image Mosaicing creates a new image by composing video frames or still images that are related, and performed by arrangement, composition and redundancy analysis of images. This paper proposes a hierarchical image mosaicing system using camera and object parameters far efficient video database construction. A tree-based image mosiacing has implemented for high-speed computation time and for construction of static and dynamic image mosaic. Camera parameters are measured by using least sum of squared difference and affine model. Dynamic object detection algorithm has proposed for extracting dynamic objects. For object extraction, difference image, macro block, region splitting and 4-split detection methods are proposed and used. Also, a dynamic positioning method is used for presenting dynamic objects and a blurring method is used for creating flexible mosaic image.

• The Journal of Korean Institute of Electromagnetic Engineering and Science
• /
• v.27 no.1
• /
• pp.69-75
• /
• 2016

In this paper, we propose a inverse synthetic aperture radar(ISAR) rotational motion compensation(RMC) method to remove residual blurring caused by non-uniform rotational motion of a target. First, a range bin having an isolated scatterer is selected. Next, polynomial phase signal in the selected range bin is estimated by using both Fourier transform(FT) and polynomial-phase transform(PPT). Finally, a new slow time variable that uniformly samples radar signal along the aspect angle directions is defined by using the estimated phase signal, and we interpolate radar signal in terms of the new time variable. As a result, rotational motion to blurr ISAR images is removed, and focused ISAR images are obtained. Simulation results using battleship model validate the robustness and effectiveness of our proposed RMC method.