• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.35 no.8C
• /
• pp.651-660
• /
• 2010

In this paper, we implement an efficient histogram-based prefiltering to compensate the illumination mismatches in regions between neighboring views. In multi-view video, such illumination disharmony can primarily occur on account of different camera location and orientation and an imperfect camera calibration. This discrepancy can cause the performance decrease of multi-view video coding(MVC) algorithm. A histogram matching algorithm can be exploited to make up for these differences in a prefiltering step. Once all camera frames of a multi-view sequence are adjusted to a predefined reference through the histogram matching, the coding efficiency of MVC is improved. However general frames of multi-view video sequence are composed of several regions with different color composition and their histogram distribution which are mutually independent of each other. In addition, the location and depth of these objects from sequeuces captured from different cameras can be different with different frames. Thus we propose a new algorithm which classify a image into several subpartitions by its depth information first and then histogram matching is performed for each region individually. Experimental results show that the compression ratio for the proposed algorithm is improved comparing with the conventional image-based algorithms.

• Journal of Korea Multimedia Society
• /
• v.17 no.2
• /
• pp.132-142
• /
• 2014

Stereo camera is drawing attention as an essential sensor for future intelligence robot system since it has the advantage of acquiring not only distance but also other additive information for an object. However, it cannot match correlated point on target image for low textured region or periodic patterned region such as wall of building or room. In this paper, we propose a stereo matching technique that increase the matching performance by fusing belief propagation stereo matching algorithm and local distance measurements of 2D-laser range finder in order to overcome this kind of limitation. The proposed technique adds laser measurements by referring quad-tree based segment information on to the local-evidence of belief propagation stereo matching algorithm, and calculates compatibility function by reflecting over-segmented information. Experimental results of the proposed method using simulation and real test images show that the distance information for some low textured region can be acquired and the discontinuity of depth information is preserved by using segmentation information.

• Journal of the Korean Institute of Telematics and Electronics S
• /
• v.36S no.10
• /
• pp.123-131
• /
• 1999

In this paper, we proposed stereo matching algorithm by dynamic programming with edges emphasized. Existing algorithms show blur generally at depth discontinuities owing to smoothness constraint and non-existence of matching pixel in occlusion regions. Also it accompanies matching error by lackness of matching information in the untextured regions. This paper defines new cost function to make up for the problems occurred to existing algorithms. It is possible through deriving matching of edges in left and right images to be carried out between edge regions anf deriving that in the other regions to be peformed between the other regions. In case of the possibility that edges can be Produced in a large amount, matching between edge information adds weight to cost function in proportion to Path distance. Proposed algorithm was applied to various images obtained by convergent camera model as well as parallel camera model. As the result, proposed algorithm showed improved performance in the aspect of matching error and processing in the occlusion regions compared to existing algorithms. Also it could improve blur especially in discontinuity regions.

• Proceedings of the Korean Information Science Society Conference
• /
• 2004.04a
• /
• pp.952-954
• /
• 2004

We've developed a Boolean circuit that finds a maximum matching in a convex bipartite graph. This circuit is designed in BC language that was created by K. Park and H. Park(1). The depth of the circuit is O(log$^2$nㆍlog b) and the size is O(bn$^3$). Our circuit gets a triple representation of a convex bipartite graph as its input and produces the maximum matching for its output. We developed some Boolean circuit design techniques that can be used for building other Boolean circuits.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.9 no.1
• /
• pp.242-259
• /
• 2015

Three-dimensional video coding is one of the main challenges restricting the widespread applications of 3D video and free viewpoint video. In this paper, a novel fractal coding algorithm with motion-vector-field-based motion estimation for depth map sequence is proposed. We firstly add pre-search restriction to rule the improper domain blocks out of the matching search process so that the number of blocks involved in the search process can be restricted to a smaller size. Some improvements for motion estimation including initial search point prediction, threshold transition condition and early termination condition are made based on the feature of fractal coding. The motion-vector-field-based adaptive hexagon search algorithm on the basis of center-biased distribution characteristics of depth motion vector is proposed to accelerate the search. Experimental results show that the proposed algorithm can reach optimum levels of quality and save the coding time. The PSNR of synthesized view is increased by 0.56 dB with 36.97% bit rate decrease on average compared with H.264 Full Search. And the depth encoding time is saved by up to 66.47%. Moreover, the proposed fractal depth map sequence codec outperforms the recent alternative codecs by improving the H.264/AVC, especially in much bitrate saving and encoding time reduction.

• Journal of IKEEE
• /
• v.28 no.1
• /
• pp.65-71
• /
• 2024

This paper proposes a neural network that can reduce the number of parameters while reducing matching errors in occluded regions to increase the accuracy of depth maps in stereo matching. Stereo matching-based object recognition is utilized in many fields to more accurately recognize situations using images. When there are many objects in a complex image, an occluded area is generated due to overlap between objects and occlusion by background, thereby lowering the accuracy of the depth map. To solve this problem, existing research methods that create context information and combine it with the cost volume or RoIselect in the occluded area increase the complexity of neural networks, making it difficult to learn and expensive to implement. In this paper, we create a depthwise seperable neural network that enhances regional feature extraction before cost volume generation, reducing the number of parameters and proposing a neural network that is robust to occlusion errors. Compared to PSMNet, the proposed neural network reduced the number of parameters by 30%, improving 5.3% in color error and 3.6% in test loss.

• Proceedings of the KIEE Conference
• /
• 1987.07b
• /
• pp.1441-1444
• /
• 1987

Extraction of dept.h information from stereo image uses the matching process between them and this requires a lot of computational time. In this paper, a matching using the feature points on the epipolar line is presented to save the computations. Feature points are obtained in both image and correlated each other. With the coordinates of the matched feature points and camera geometry, the position and depth informations are identified.

• Journal of Broadcast Engineering
• /
• v.17 no.3
• /
• pp.437-446
• /
• 2012

A face detection algorithms using two-dimensional (2-D) intensity or color images have been studied for decades. Recently, with the development of low-cost range sensor, three-dimensional (3-D) information (i.e., depth image that represents the distance between a camera and objects) can be easily used to reliably extract facial features. Most people have a similar pattern of 3-D facial structure. This paper proposes a face detection method using intensity and depth images. At first, adaboost algorithm using intensity image classifies face and nonface candidate regions. Each candidate region is divided into $5{\times}5$ blocks and depth values are averaged in each block. Then, $5{\times}5$ block rank pattern is constructed by sorting block averages of depth values. Finally, candidate regions are classified as face and nonface regions by matching the constructed depth map based block rank patterns and a template pattern that is generated from training data set. For template matching, the $5{\times}5$ template block rank pattern is prior constructed by averaging block ranks using training data set. The proposed algorithm is tested on real images obtained by Kinect range sensor. Experimental results show that the proposed algorithm effectively eliminates most false positives with true positives well preserved.

• Journal of the Korean Institute of Telematics and Electronics B
• /
• v.30B no.9
• /
• pp.64-75
• /
• 1993

This paper proposes a recovering technique of the elevation map by stereo modeling of the aerial image sequence which is transformed based on the aircraft situation. The area-based stereo matching method is simulated and the various parameters are experimentally chosen. In a depth extraction step, the depth is determined by solving the vector equation. The equation is suitable for stereo modeling of aerial images which do not satisfy the epipolar constraint. Also, the performance of the conventional feature-based matching scheme is compared. Finally, techniques analyzing the accuracy of the recovered elevation map (REM) are described. The analysis includes the error estimation for both height and contour lines, where the accuracy is based on the measurements of deviations from the estimates obtained manually. The experimental results show the efficiency of the proposed technique.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.6 no.10
• /
• pp.2663-2678
• /
• 2012

Due to sharp depth transition, big holes may be found in the novel view that is synthesized by depth-image-based rendering (DIBR). A hole-filling method based on disparity map is proposed. One important aspect of the method is that the disparity map of destination image is used for hole-filling, instead of the depth image of reference image. Firstly, the big hole detection based on disparity map is conducted, and the start point and the end point of the hole are recorded. Then foreground pixels and background pixels are distinguished for hole-dilating according to disparity map, so that areas with matching errors can be determined and eliminated. In addition, parallaxes of pixels in the area with holes and matching errors are changed to new values. Finally, holes are filled with background pixels from reference image according to these new parallaxes. Experimental results show that the quality of the new view after hole-filling is quite well; and geometric distortions are avoided in destination image, in contrast to the virtual view generated by depth-smoothing methods and image inpainting methods. Moreover, this method is easy for hardware implementation.