• KIPS Transactions on Software and Data Engineering
• /
• v.12 no.5
• /
• pp.217-228
• /
• 2023

Video instance segmentation is an intelligent visual task with high complexity because it not only requires object instance segmentation for each image frame constituting a video, but also requires accurate tracking of instances throughout the frame sequence of the video. In special, human instance segmentation in drama videos has an unique characteristic that requires accurate tracking of several main characters interacting in various places and times. Also, it is also characterized by a kind of the class imbalance problem because there is a significant difference between the frequency of main characters and that of supporting or auxiliary characters in drama videos. In this paper, we introduce a new human instance datatset called MHIS, which is built upon drama videos, Miseang, and then propose a novel video data augmentation method, CDVA, in order to overcome the data imbalance problem between character classes. Different from the previous video data augmentation methods, the proposed CDVA generates more realistic augmented videos by deciding the optimal location within the background clip for a target human instance to be inserted with taking rich spatio-temporal context embedded in videos into account. Therefore, the proposed augmentation method, CDVA, can improve the performance of a deep neural network model for video instance segmentation. Conducting both quantitative and qualitative experiments using the MHIS dataset, we prove the usefulness and effectiveness of the proposed video data augmentation method.

• The Transactions of the Korea Information Processing Society
• /
• v.6 no.4
• /
• pp.1081-1089
• /
• 1999

Volume data modeling is concerned with finding a mathematical function which represents the relationship implied by the 3D data. Modeling a volume data geometrically can visualize a volume data using surface graphics without voxelization. It has many merits in that it is fast and requires little memory. We proposes, a method based on wavelet transformation and tetrahedrization. we implement a prototype system based on the proposed method. Last, we evaluated the proposed method comparing it with marching cube algorithm. the evaluation results show that though the proposed method uses only 13% of the volume data, the images generated is as good as the images generated by the marching cubes algorithm.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.18 no.1
• /
• pp.103-108
• /
• 2014

The flash memory which is exploited on hand-held devices such as smart phones is a non-volatile storage and has the benefit that it can store mass data at a small sized chip. To process queries on the mass data stored in the flash memory, the index scheme should be exploited. However, since the write operation of the flash memory is slower than the read operation and the overwrite is not supported, it is required to reevaluate the performance of the index and find out the drawbacks. In this paper, we evaluate the performance of a space division typed index scheme on the flash memory. To do this, we implement the fixed grid file and measure the average speeds of the query and update processing on a various condition and compare the value of the flash memory with that of the magnetic disk.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.40 no.1
• /
• pp.63-71
• /
• 2003

This paper describes a color image segmentation algorithm based on region merging using hue difference as a restrictive condition. The proposed algorithm using mathematical morphology and a modified watershed algorithm does over-segmentation in the RGB space to preserve contour information of regions. Then, the segmentation result of color image is acquired by repeated region merging using hue differences as a restrictive condition. This stems from human visual system based on hue, saturation, and intensity. Hue difference between two regions is used as a restrictive condition for region merging because it becomes more important factor than color difference if intensity is not low. Simulation results show that the proposed color image segmentation algorithm provides efficient segmentation results with the predefined number of regions for various color images.

• The Journal of the Acoustical Society of Korea
• /
• v.37 no.6
• /
• pp.489-498
• /
• 2018

In this paper, a target detection method based on beamspace-domain multichannel nonnegative matrix factorization is studied when an echo of continuous-wave ping is received from a low-Doppler target in reverberant environment. If the receiver of the continuous-wave active sonar moves, the frequency range of the reverberation is broadened due to the Doppler effect, so the low-Doppler target echo is interfered by the reverberation in this case. The developed algorithm analyzes the multichannel spectrogram of the received signal into frequency bases, time bases, and beamformer gains using the beamspace-domain multichannel nonnnegative matrix factorization, then the algorithm estimates the frequency, time, and bearing of target echo by choosing a proper basis. To analyze the performance of the developed algorithm, simulations were performed in various signal-to-reverberation conditions. The results show that the proposed algorithm can estimate the frequency, time, and bearing, but the performance was degraded in the low signal-to-reverberation condition. It is expected that modifying the selection algorithm of the target echo basis can enhance the performance according to the simulation results.

• Journal of the Korean Society of Surveying, Geodesy, Photogrammetry and Cartography
• /
• v.28 no.6
• /
• pp.627-636
• /
• 2010

Image segmentation technique is becoming increasingly important in the field of remote sensing image analysis in areas such as object oriented image classification to extract object regions of interest within images. This paper presents a new method for image segmentation to consider spectral and spatial information of high resolution satellite image. Firstly, the initial seeds were automatically selected using local variation of multi-spectral edge information. After automatic selection of significant seeds, a segmentation was achieved by applying MSRG which determines the priority of region growing using information drawn from similarity between the extracted each seed and its neighboring points. In order to evaluate the performance of the proposed method, the results obtained using the proposed method were compared with the results obtained using conventional region growing and watershed method. The quantitative comparison was done using the unsupervised objective evaluation method and the object-based classification result. Experimental results demonstrated that the proposed method has good potential for application in the object-based analysis of high resolution satellite images.

• Journal of KIISE:Software and Applications
• /
• v.32 no.7
• /
• pp.625-635
• /
• 2005

In this paper, we propose a hybrid approach for segmenting the lungs efficiently and automatically in chest CT images. The proposed method consists of the following three steps. first, lungs and airways are extracted by two- and three-dimensional automatic seeded region growing and connected component labeling in low-resolution. Second, trachea and large airways are delineated from the lungs by two-dimensional morphological operations, and the left and right lungs are identified by connected component labeling in low-resolution. Third, smooth and accurate lung region borders are obtained by refinement based on image subtraction. In experiments, we evaluate our method in aspects of accuracy and efficiency using 10 chest CT images obtained from 5 patients. To evaluate the accuracy, we Present results comparing our automatic method to manually traced borders from radiologists. Experimental results show that proposed method which use connected component labeling in low-resolution reduce processing time by 31.4 seconds and maximum memory usage by 196.75 MB on average. Our method extracts lung surfaces efficiently and automatically without additional processing like hole-filling.

• The Transactions of the Korea Information Processing Society
• /
• v.6 no.7
• /
• pp.1849-1857
• /
• 1999

We proposed the RPA (Recursive Partition Averaging) method in order to improve the storage requirement and classification rate of the Memory Based Reasoning. This algorithm recursively partitions the pattern space until each hyperrectangle contains only those patterns of the same class, then it computes the average values of patterns in each hyperrectangle to extract a representative. Also we have used the mutual information between the features and classes as weights for features to improve the classification performance. The proposed algorithm used 30~90% of memory space that is needed in the k-NN (k-Nearest Neighbors) classifier, and showed a comparable classification performance to the k-NN. Also, by reducing the number of stored patterns, it showed an excellent result in terms of classification time when we compare it to the k-NN.

• Journal of Korea Spatial Information System Society
• /
• v.6 no.2 s.12
• /
• pp.23-37
• /
• 2004

Location Based Services(LBSs) in mobile environments become very popular recently. For efficient LBSs, spatial database management systems must need a spatial indexing scheme such as R-trees in order to manage the huge spatial database. However, it may need unnecessary disk accesses since it needs to access objects which are not actually concerned to user's location-based queries. In this paper, to support the location-based queries efficiently, we propose a CLR-tree(Cell Leveling R-tree) in which a dynamic cell is built up within the minimum bounding rectangle of R-trees' node. The cell level of nodes is compared with the query's cell level in location-based query processing and determines the minimum search space. Also, we propose the insertion, split, deletion, and search algorithms for CRL-trees. From the experimental results, we see that a CLR-tree is able to decrease $5{\sim}20%$ of disk accesses from those of R-trees. So, a CLR-tree can be used for fast accessing spatial objects to user's location-based queries in LBSs.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.26 no.7B
• /
• pp.968-974
• /
• 2001

최근 정보통신기술의 발전과 함께 영상매체의 급속한 증가로 영상의 효율적인 관리와 검색의 필요성이 요구되면서 내용기반 영상검색이 핵심기술로 대두되고 있다. 내용기반 영상검색에서 영상의 특징을 표현하기 위해 색상 히스토그램을 많이 사용하고 있으나, 색상만을 고려하는 것은 많은 단점을 지니고 있으므로 본 논문에서는 먼저 순차영역분할(sequential clustering)기법을 도입하여 영역을 분할하며, 분할된 영역의 색상평균값과 영역의 중심점으로부터의 거리 히스토그램을 영상의 특징으로 구하여 이를 비교함으로써 색상과 공간정보를 함께 고려하는 방법을 제안한다. 제안된 방법의 특성의 수가 18개로 타 방법보다 매우 작은 저장공간을 가지면서도 동시에 검색효율이 8.5% 이상 개선되었다. Precision 대 Recall에서도 각 질의영상에서 대부분의 Recall 값에서 제안한 방법의 우수함이 확인되었으며, 시각적으로도 양호한 검색결과를 얻을 수 있었다.