• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.18 no.9
• /
• pp.2605-2625
• /
• 2024

Siamese-based segmentation and tracking algorithms improve accuracy and stability for video object segmentation and tracking tasks simultaneously. Although effective, variability in target appearance and background clutter can still affect segmentation accuracy and further influence the performance of tracking. In this paper, we present a memory propagation-based target-aware and mask-attention decision network for robust object segmentation and tracking. Firstly, a mask propagation-based attention module (MPAM) is constructed to explore the inherent correlation among image frames, which can mine mask information of the historical frames. By retrieving a memory bank (MB) that stores features and binary masks of historical frames, target attention maps are generated to highlight the target region on backbone features, thus suppressing the adverse effects of background clutter. Secondly, an attention refinement pathway (ARP) is designed to further refine the segmentation profile in the process of mask generation. A lightweight attention mechanism is introduced to calculate the weight of low-level features, paying more attention to low-level features sensitive to edge detail so as to obtain segmentation results. Finally, a mask fusion mechanism (MFM) is proposed to enhance the accuracy of the mask. By utilizing a mask quality assessment decision network, the corresponding quality scores of the "initial mask" and the "previous mask" can be obtained adaptively, thus achieving the assignment of weights and the fusion of masks. Therefore, the final mask enjoys higher accuracy and stability. Experimental results on multiple benchmarks demonstrate that our algorithm performs outstanding performance in a variety of challenging tracking tasks.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.26 no.3
• /
• pp.381-388
• /
• 2022

In instance segmentation, Mask-RCNN is mostly used as a base model. Increasing the performance of Mask-RCNN is meaningful because it affects the performance of the derived model. Mask-RCNN has a transform module for unifying size of input images. In this paper, to improve the Mask-RCNN, we apply deep-learning-based ASSR to the resizing part in the transform module and inject calculated scale information into the model using IM(Integration Module). The proposed IM improves instance segmentation performance by 2.5 AP higher than Mask-RCNN in the COCO dataset, and in the periment for optimizing the IM location, the best performance was shown when it was located in the 'Top' before FPN and backbone were combined. Therefore, the proposed method can improve the performance of models using Mask-RCNN as a base model.

• Proceedings of the Korean Institute of Electrical and Electronic Material Engineers Conference
• /
• 2004.04b
• /
• pp.97-100
• /
• 2004

We have grown GaN layers with in-situ SiN mask by metal organic chemical vapor deposition (MOCVD) and study the characteristic of the GaN layer. We have changed the deposition time of SiN mask from 45s to 5min and obtain th optimum condition in 45s. The PL intensity of GaN with SiN mask improved 2 times to that without SiN mask and the carrier concentraion increased from $3.5{\times}10^{16}cm^{-3}$ to $1.8{\times}10^{17}cm^{-3}$. We have thus shown that the SiN mask improved significantly the optical properties of the GaN layer.

• Proceedings of the Korean Society for Language and Information Conference
• /
• 1998.02a
• /
• pp.283-294
• /
• 1998

• Journal of Biomedical Engineering Research
• /
• v.16 no.2
• /
• pp.239-248
• /
• 1995

The method of increasing signal to noise ratio (SNR) in a Hadamard transform spectrometer (HTS) is multiplexing. The multiplexing is executed by a mask. Conventional masks are mechanical or electro-optical. A mechanical mask has disadvantages of jamming and misalignment. A stationary electro-optical mask has a disadvantage of information losses caused by spacers which partition mask elements. In this paper, a mixed-concept electro-optical mask (MCEOM) is developed by expanding the length of a spacer to that of lon-off mask element. An MCEOM is operated by stepping a movable mask. 2N measurements are required for N spectrum estimates. The average mean square error (AMSE) using MCEQM is equal to that using a stationary electro-optical mask without spacers for large N. The cost of manufacturing an MCEOM is lower than that of producing a conventional electro-optical mask because an MCEOM needs only (N + 1)/2 on-off mask elements whereas the con¬ventional electro-optical mask needs N on-off mask elements. There are no information losses in the spectrometers having an MCEOM.

• Journal of information and communication convergence engineering
• /
• v.21 no.1
• /
• pp.17-23
• /
• 2023

Recently, several high-performance instance segmentation models have used the Mask R-CNN model as a baseline, which reached a historical peak in instance segmentation in 2017. There are numerous derived models using the Mask R-CNN model, and if the performance of Mask R-CNN is improved, the performance of the derived models is also anticipated to improve. The Mask R-CNN uses interpolation to adjust the image size, and the input differs depending on the interpolation method. Therefore, in this study, the performance change of Mask R-CNN was compared when various interpolation methods were applied to the transform layer to improve the performance of Mask R-CNN. To train and evaluate the models, this study utilized the PennFudan and Balloon datasets and the AP metric was used to evaluate model performance. As a result of the experiment, the derived Mask R-CNN model showed the best performance when bicubic interpolation was used in the transform layer.

• Journal of the Korean Institute of Telematics and Electronics A
• /
• v.30A no.9
• /
• pp.46-58
• /
• 1993

Owing to the nature of locality in mask layouts, it appears that most mask layout problems can be solved by transforming a part of the given mask layout into a better layout segment continuously toward a global suboptimal solution. This notion of local transformation addresses major weak points of existing mask layout processing systems, which lack both extensibility and unifiability. This paper attempts to elaborate upon developing a new rule-based mask layout transformation system wherein most of the mask layout problems can be solved under the unified framework of local mask layout transformation. The rule-based mask layout transformation system is applicable to various mask layout problems such as net extraction, mask layout compaction, mask layout editing, and design rule checking. The experimental results show that the rule-based expert system approach is an efficient means of solving those mask layout problems, and thus confronting major drawbacks of existing layout processing systems.

• Journal of the Korean Institute of Telematics and Electronics
• /
• v.26 no.8
• /
• pp.1247-1256
• /
• 1989

In this paper, we propose the dynamic scene segmentation algorithm using a cross mask and edge information. This method, a combination of the conventioanl feature-based and pixel-based approaches, uses edges as features and determines moving pixels, with a cross mask centered on each edge pixel, by computing similarity measure between two consecutive image frames. With simple calcualtion the proposed method works well for image consisting of complex background or several moving objects. Also this method works satisfactorily in case of rotaitional motion.

• Journal of information and communication convergence engineering
• /
• v.13 no.4
• /
• pp.280-285
• /
• 2015

Double random phase encryption (DRPE) is one of the well-known optical encryption techniques, and many techniques with DRPE have been developed for information security. However, most of these techniques may not solve the fundamental security problem caused by using fixed phase masks for DRPE. Therefore, in this paper, we propose a key phase mask updating scheme for DRPE to improve its security, where a spatial light modulator (SLM) is used to implement key phase mask updating. In the proposed scheme, updated key data are obtained by using previous image data and the first phase mask used in encryption. The SLM with the updated key is used as the second phase mask for encryption. We provide a detailed description of the method of encryption and decryption for a DRPE system using the proposed key updating scheme, and simulation results are also shown to verify that the proposed key updating scheme can enhance the security of the original DRPE.

• Journal of the Korea Society of Computer and Information
• /
• v.16 no.3
• /
• pp.27-34
• /
• 2011

In this paper, the novel approach for image contrast enhancement is introduced. The method is based on the unsharp mask and directional information of images. Since the unsharp mask techniques give better visual quality than the conventional sharpening mask, there are much works on image enhancement using unsharp masks. The proposed algorithm decomposes the image to several blocks and extracts directional information using DCT. From the geometric properties of the block, each block is labeled as appropriate type and processed by adaptive unsharp mask. The masking process is skipped at the flat area to reduce the noise artifact, but at the texture and edge area, the adaptive unsharp mask is applied to enhance the image contrast based on the edge direction. Experiments show that the proposed algorithm produces the contrast enhanced images with superior visual quality, suppressing the noise effects and enhancing edge at the same time.