• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.16 no.3
• /
• pp.877-893
• /
• 2022

With the development of deep learning, face inpainting has been significantly enhanced in the past few years. Although image inpainting framework integrated with generative adversarial network or attention mechanism enhanced the semantic understanding among facial components, the issues of reconstruction on corrupted regions are still worthy to explore, such as blurred edge structure, excessive smoothness, unreasonable semantic understanding and visual artifacts, etc. To address these issues, we propose a Learnable Structure Knowledge of Fusion Network (LSK-FNet), which learns a prior knowledge by edge generation network for image inpainting. The architecture involves two steps: Firstly, structure information obtained by edge generation network is used as the prior knowledge for face inpainting network. Secondly, both the generated prior knowledge and the incomplete image are fed into the face inpainting network together to get the fusion information. To improve the accuracy of inpainting, both of gated convolution and region normalization are applied in our proposed model. We evaluate our LSK-FNet qualitatively and quantitatively on the CelebA-HQ dataset. The experimental results demonstrate that the edge structure and details of facial images can be improved by using LSK-FNet. Our model surpasses the compared models on L1, PSNR and SSIM metrics. When the masked region is less than 20%, L1 loss reduce by more than 4.3%.

• Journal of the Korean Geographical Society
• /
• v.38 no.5
• /
• pp.667-685
• /
• 2003

To address the requirements of gap analysis for species protection, as well as the needs of state and federal agencies for detailed digital land cover, a 43-class map at the vegetation alliance level was created for the state of Kansas using multi-temporal Thematic Mapper imagery. The mapping approach included the use of three-date multi-seasonal imagery, a two-stage classification approach that first masked out cropland areas using unsupervised classification and then mapped natural vegetation with supervised classification, visualization techniques utilizing a map of small multiples and field experts, and extensive use of ancillary data in post-hoc processing. Accuracy assessment was conducted at three levels of generalization (Anderson Level I, vegetation formation, and vegetation alliance) and three cross-tabulation approaches. Overall accuracy ranged from 51.7% to 89.4%, depending on level of generalization, while accuracy figures for individual alliance classes varied by area covered and level of sampling.

• Asian Pacific Journal of Cancer Prevention
• /
• v.14 no.12
• /
• pp.7731-7736
• /
• 2013

Aim: To compare the agreement of screening breast mammography plus ultrasound and reviewed mammography alone in asymptomatic women. Materials and Methods: All breast imaging data were obtained for women who presented for routine medical checkup at National Cancer Institute (NCI), Thailand from January 2010 to June 2013. A radiologist performed masked interpretations of selected mammographic images retrieved from the computer imaging database. Previous mammography, ultrasound reports and clinical data were blinded before film re-interpretation. Kappa values were calculated to assess the agreement between BIRADS assessment category and BIRADS classification of density obtained from the mammography with ultrasound in imaging database and reviewed mammography alone. Results: Regarding BIRADS assessment category, concordance between the two interpretations were good. Observed agreement was 96.1%. There was moderate agreement in which the Kappa value was 0.58% (95%CI; 0.45, 0.87). The agreement of BI-RADS classification of density was substantial, with a Kappa value of 0.60 (95%CI; 0.54, 0.66). Different results were obtained when a subgroup of patients aged ${\geq}60$ years were analyzed. In women in this group, observed agreement was 97.6%. There was also substantial agreement in which the Kappa value was 0.74% (95%CI; 0.49, 0.98). Conclusions: The present study revealed that concordance between mammography plus ultrasound and reviewed mammography alone in asymptomatic women is good. However, there is just moderate agreement which can be enhanced if age-targeted breast imaging is performed. Substantial agreement can be achieved in women aged ${\geq}60$. Adjunctive breast ultrasound is less important in women in this group.

• Journal of Astronomy and Space Sciences
• /
• v.27 no.4
• /
• pp.401-406
• /
• 2010

We have measured and analyzed the dark data of two PICNIC IR arrays (P574 and P560) obtained through the Cosmic Infrared Background ExpeRiment (CIBER). First, we identified three types of bad pixels: the cold, the hot, and the transient, which are figured in total as 0.06% for P574 and 0.19% for P560. Then, after the bad pixels were masked, we determined the dark noise to be 20.5 ${\pm}$ 0.05 $e^-$ and 16.1 ${\pm}$ 0.05 $e^-$, and the dark current to be 0.6 ${\pm}$ 0.05 $e^-$/sec and 0.7 ${\pm}$ 0.05 $e^-$/sec for P574 and P560, respectively. Finally, we discussed glitches and readout modes for a future mission.

• Journal of Broadcast Engineering
• /
• v.18 no.2
• /
• pp.311-318
• /
• 2013

This paper deals with a noise reduction algorithm which uses the binary masking in the time-frequency domain. To improve speech intelligibility in noise, noise-masked speech is decomposed into time-frequency units and mask "0" is assigned to masker-dominant region removing time-frequency units where noise is dominant compared to speech. In the previous research, Gaussian mixture models were used to classify the speech-dominant region and noise-dominant region which correspond to mask "1" and mask "0", respectively. In each frequency band, data were collected and trained to build the Gaussian mixture models and detection procedure is performed to the test data where each time-frequency unit belongs to speech-dominant region or noise-dominant region. In this paper, we consider the correlation of masks in the frequency domain and propose a post-processing method which exploits the Viterbi algorithm.

• Journal of the Korean Society for Nondestructive Testing
• /
• v.20 no.6
• /
• pp.545-554
• /
• 2000

Compton X-ray backscatter technique has been used to quantitatively assess the impact damage in quasi-isotropic laminated composites and to obtain a cross-sectional profile of impact-damaged laminated composites from the density variation of the cross section. An adaptive filter is applied to the Compton backscattering data for the reconstruction and noise reduction from many sources including quantum noise, especially when the SNR(signal-to-noise ratio) of the image is relatively low. A nonlinear reconstruction model is also proposed to overcome distortion of the Compton backscatter image due to attenuation effects, beam hardening, and irregular distributions of the fibers and the matrix in composites. Delaminations masked or distorted by the first few delaminations near the front surface are detected and characterized both in width and location, by application of an error minimization algorithm.

• Journal of Korean Ophthalmic Optics Society
• /
• v.11 no.3
• /
• pp.241-247
• /
• 2006

Purpose : To define the distribution of normal corneal topographic patterns of Korean college students and compare them with previously published western data, Bogan et al's study. Methods : Computerized corneal topography was performed 220eyes normal subjects using CTK-922(Topography, swiss made). Mean age of the subjects was 23.2 yr.(range 19 to 57 yr.). The color-coded videokeratographs were classified by a masked observer according to the Bogan et al's classification such as round, oval, symmetric bow tie, asymmetric bow tie, and irregular. Results: The results revealed 57 eyes(25.9%) had round, 14 (6.4%) oval , 41(18.6%)symmetric bow tie, 78(35.5%) asymmetric bow tie, and 30(13.6%) irregular pattern. Conclusions : Our results of topographic patterns show the tendency of more irregular and less round topographic pattern in Korean college students, as compared to that of western adults.

• Journal of Fisheries and Marine Sciences Education
• /
• v.12 no.2
• /
• pp.191-198
• /
• 2000

An experiment was carried out to obtain the fundamental data on the auditory thresholds of fishes for catching method using audible frequency sound, the auditory thresholds of dusky spinefoot Siganus fuscescens(Houttuyn) were measured in the presence of masking noise in the spectrum level range of 74 - 83dB re $1{\mu}Pa/{\sqrt{Hz}}$ by heartbeat conditioning technique using pure tones coupled with a delayed electric shock. The auditory critical ratios were about 23 - 34dB at measurement frequency range. The ratio increased almost linearly with increasing frequency from 200 to 500Hz. The noise spectrum level at the start of masking was about 61 - 73dB within the measurement frequency range. This suggests that hearing of dusky spinefoot is masked in the natural environment with the noise spectrum level above 70dB. The sound pressure level of which the signal sound of 100Hz is recognized by dusky spinefoot under the white noise of 70dB is above 98dB and the critical ratio of them is above 23dB.

• Korean Journal of Fisheries and Aquatic Sciences
• /
• v.34 no.2
• /
• pp.151-155
• /
• 2001

In order to obtain the fundamental data on the auditory thresholds of fishes for marine ranching, the auditory thresholds of black rockfish Sebastes inermis were measured in the presence of masking noise in the spectrum level range of $73\~83$ dB (0 dB re $1{\mu}Pa/\sqrt{Hz}$) with a classical cardiac conditioning technique. Critical ratios were about $28\~34$ dB at $80\~300$ Hz and $47\~52$ dB at $500\~800$ Hz. The ratio increased almost linearly with increasing frequency to 500 Hz. The noise spectrum level at the start of masking was about 70 dB within the frequency range of $80\~800$ Hz excepting 65 dB at 300 Hz. It means that hearing of the black rockfish is masked in the natural environment with the noise spectrum level above 65 dB. The sound pressure level of $200\~300$ Hz recognized by black rockfish was above 96 dB under the ambient noise and the critical ratio of them was above 26 dB.

• Transactions of the Korean Society of Automotive Engineers
• /
• v.25 no.1
• /
• pp.92-102
• /
• 2017

For an automotive surveillance radar system, fast-chirp train based FMCW (Frequency Modulated Continuous Wave) radar is a very effective method, because clutter and moving targets are easily separated in a 2D range-velocity map. However, pedestrians with low echo signals may be masked by strong clutter in actual field. To address this problem, we proposed in the previous work a clutter cancellation and moving target indication algorithm using the coherent phase method. In the present paper, we initially composed the test set-up using a 24 GHz FMCW transceiver and a real-time data logging board in order to verify this algorithm. Next, we created two indoor test environments consisting of moving human and stationary targets. It was found that pedestrians and strong clutter could be effectively separated when the proposed method is used. We also designed and implemented these algorithms in FPGA (Field Programmable Gate Array) in order to analyze the hardware and time complexities. The results demonstrated that the complexity overhead was nearly zero compared to when the typical method was used.