• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.18 no.1
• /
• pp.15-29
• /
• 2024

A fetoscope is an optical endoscope, which is often applied in fetoscopic laser photocoagulation to treat twin-to-twin transfusion syndrome. In an operation, the clinician needs to observe the abnormal placental vessels through the endoscope, so as to guide the operation. However, low-quality imaging and narrow field of view of the fetoscope increase the difficulty of the operation. Introducing an accurate placental vessel segmentation of fetoscopic images can assist the fetoscopic laser photocoagulation and help identify the abnormal vessels. This study proposes a method to solve the above problems. A novel encoder-decoder network with a dual-path structure is proposed to segment the placental vessels in fetoscopic images. In particular, we introduce a channel attention mechanism and a continuous convolution structure to obtain multi-scale features with their weights. Moreover, a switching connection is inserted between the corresponding blocks of the two paths to strengthen their relationship. According to the results of a set of blood vessel segmentation experiments conducted on a public fetoscopic image dataset, our method has achieved higher scores than the current mainstream segmentation methods, raising the dice similarity coefficient, intersection over union, and pixel accuracy by 5.80%, 8.39% and 0.62%, respectively.

• Journal of Broadcast Engineering
• /
• v.14 no.4
• /
• pp.450-462
• /
• 2009

So far, many methods for segmenting single images or video have been proposed, but few methods have dealt with multiple images with analogous content. These images, which we term consistent scene images, include concurrent images of a scene and gathered images of a similar foreground, and may be collectively utilized to describe a scene or as input images for multi-view stereo. In this paper, we present a method to segment these images with minimum user input, specifically, manual segmentation of one image, by iteratively propagating information via multi-level cues with adaptive confidence depending on the nature of the images. Propagated cues are used as the bases to compute multi-level potentials in an MRF framework, and segmentation is done by energy minimization. Both cues and potentials are classified as low-, mid-, and high- levels based on whether they pertain to pixels, patches, and shapes. A major aspect of our approach is utilizing mid-level cues to compute low- and mid- level potentials, and high-level cues to compute low-, mid-, and high- level potentials, thereby making use of inherent information. Through this process, the proposed method attempts to maximize the amount of both extracted and utilized information in order to maximize the consistency of the segmentation. We demonstrate the effectiveness of the proposed method on several sets of consistent scene images and provide a comparison with results based only on mid-level cues [1].

• The Journal of the Korea Contents Association
• /
• v.10 no.1
• /
• pp.1-9
• /
• 2010

In this paper we present a shading technique for realistic rendering of the surfaces with irregular and complex textures using a single photograph. So far, most works have been using many photographs or special photographing equipment to render the surfaces with irregular and complex textures as well as dividing texture regions manually. We present an automatic selection method of the region segmentation techniques according to properties of materials. As our technique produces a reflectance model and the approximated Bidirectional Reflection Distribution Function(BRDF) parameters, it allows the recovery of the photometric properties of diffuse, specular, isotropic or anisotropic textured objects. Also it make it possible to present several synthetic images with novel lighting conditions and views.

• Journal of Biomedical Engineering Research
• /
• v.26 no.2
• /
• pp.73-86
• /
• 2005

We need to flatten the brain cortex to smooth surface, sphere, or 2D plane in order to view the buried sulci. The rendered 3D surface of the segmented white matter and gray matter does not have the topology of a sphere due to the partial volume effect and segmentation error. A surface without correct topology may lead to incorrect interpretation of local structural relationships and prevent cortical unfolding. Although some algorithms try to correct topology, they require heavy computation and fail to follow the deep and narrow sulci. This paper proposes a method that corrects topology of the rendered surface fast, accurately, and automatically. The proposed method removes fractions beside the main surface, fills cavities in the inside of the main surface, and removes handles in the surface. The proposed method to remove handles has three-step approach. Step 1 performs smoothing operation on the rendered surface. In Step 2, vertices of sphere are gradually deformed to the smoothed surfaces and finally to the boundary of the segmented white matter and gray matter. The Step 2 uses multi-resolutional approach to prevent the deep sulci from geometrical intersection. In Step 3, 3D binary image is constructed from the deformed sphere of Step 2 and 3D surface is regenerated from the 3D binary image to remove intersection that may happen. The experimental results show that the topology is corrected while principle sulci and gyri are preserved and the computation amount is acceptable.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.27 no.1A
• /
• pp.80-88
• /
• 2002

Visual media(image, graphic, and video) processing poses challenge from several perpectives, specifically from the point of view of real-time implementation and scalability. There have been several approaches to obtain speedups to meet the computing demands in multimedia processing ranging from media processors to special purpose implementations. A variety of parallel processing strategies are adopted in these implementations in order to achieve the required speedups. We have investigated a parallel processing system for improving the processing speed o f visual media related applications. The parallel processing system we proposed is similar to a pipelined memory stystem(MAMS). The multi-access memory system is made up of m memory modules and a memory controller to perform parallel memory access with a variety of combinations of 1${\times}$pq, pq${\times}$1, and p${\times}$q subarray, which improves both cost and complexity of control. Facial recognition, Phong shading, and automatic segmentation of moving object in image sequences are some that have been applied to the parallel processing system and resulted in faithful processing speed. This paper describes the parallel processing systems for the speedup and its utilization to three time-consuming applications.

• The KIPS Transactions:PartB
• /
• v.18B no.6
• /
• pp.333-340
• /
• 2011

The Human-Computer Interaction(HCI) is a study of the method for interaction between human and computers that merges the ergonomics and the information technology. The intelligent space, which is a part of the HCI, is an important area to provide effective user interface for the disabled, who are alienated from the information-oriented society. In the intelligent space for the disabled, the method supporting information depends on types of disability. In this paper, we only support the hearing-impaired. It is material to the gaze direction detection method because it is very efficient information provide method to present information on gazing direction point, except for the information provide location perception method through directly contact with the hearing-impaired. We proposed the gaze direction detection method must be necessary in order to provide the residence life application to the hearing-impaired like this. The proposed method detects the region of the user from multi-view camera images, generates candidates for directions of gaze for horizontal and vertical from each camera, and calculates the gaze direction of the user through the comparison with the size of each candidate. In experimental results, the proposed method showed high detection rate with gaze direction and foot sensing rate with user's position, and showed the performance possibility of the scenario for the disabled.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2003.05a
• /
• pp.165-170
• /
• 2003

This paper presents an integrated approach to unconstrained face recognition in arbitrary scenes. The front end of the system comprises of a scale and pose tolerant face detector. Scale normalization is achieved through novel combination of a skin color segmentation and log-polar mapping procedure. Principal component analysis is used with the multi-view approach proposed in[10] to handle the pose variations. For a given color input image, the detector encloses a face in a complex scene within a circular boundary and indicates the position of the nose. Next, for recognition, a radial grid mapping centered on the nose yields a feature vector within the circular boundary. As the width of the color segmented region provides an estimated size for the face, the extracted feature vector is scale normalized by the estimated size. The feature vector is input to a trained neural network classifier for face identification. The system was evaluated using a database of 20 person's faces with varying scale and pose obtained on different complex backgrounds. The performance of the face recognizer was also quite good except for sensitivity to small scale face images. The integrated system achieved average recognition rates of 87% to 92%.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.7 no.6
• /
• pp.1312-1317
• /
• 2003

This paper presents an integrated approach to unconstrained face recognition in arbitrary scenes. The front end of the system comprises of a scale and pose tolerant face detector. Scale normalization is achieved through novel combination of a skin color segmentation and log-polar mapping procedure. Principal component analysis is used with the multi-view approach proposed in[10] to handle the pose variations. For a given color input image, the detector encloses a face in a complex scene within a circular boundary and indicates the position of the nose. Next, for recognition, a radial grid mapping centered on the nose yields a feature vector within the circular boundary. As the width of the color segmented region provides an estimated size for the face, the extracted feature vector is scale normalized by the estimated size. The feature vector is input to a trained neural network classifier for face identification. The system was evaluated using a database of 20 person's faces with varying scale and pose obtained on different complex backgrounds. The performance of the face recognizer was also quite good except for sensitivity to small scale face images. The integrated system achieved average recognition rates of 87% to 92%.