• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2022.10a
• /
• pp.6-8
• /
• 2022

Many systems that use images through object recognition using deep learning have provided various solutions beyond the existing methods. Many studies have proven its usability, and the actual control system shows the possibility of using it to make people's work more convenient. Many studies have proven its usability, and actual control systems make human tasks more convenient and show possible. However, with hardware-intensive performance, the development of models is facing some limitations, and the ease with the use and additional utilization of many unupdated models is falling. In this paper, we propose how to increase utilization and accuracy by providing additional information on the emotional regions of colors and objects by utilizing learning and weights from HSV color histograms of local image data recognized after conventional stereotyped object recognition results.

• Investigative Magnetic Resonance Imaging
• /
• v.23 no.4
• /
• pp.328-340
• /
• 2019

Purpose: To generate phase images with free of motion-induced artifact and susceptibility-induced distortion using 3D radial ultrashort TE (UTE) MRI. Materials and Methods: The field map was theoretically derived by solving Laplace's equation with appropriate boundary conditions, and used to simulate the image distortion in conventional spin-warp MRI. Manufacturer's 3D radial imaging sequence was modified to acquire maximum number of radial spokes in a given time, by removing the spoiler gradient and sampling during both rampup and rampdown gradient. Spoke direction randomly jumps so that a readout gradient acts as a spoiling gradient for the previous spoke. The custom raw data was reconstructed using a homemade image reconstruction software, which is programmed using Python language. The method was applied to a phantom and in-vivo human brain and abdomen. The performance of UTE was compared with 3D GRE for phase mapping. Local phase mapping was compared with T₂* mapping using UTE. Results: The phase map using UTE mimics true field-map, which was theoretically calculated, while that using 3D GRE revealed both motion-induced artifact and geometric distortion. Motion-free imaging is particularly crucial for application of phase mapping for abdomen MRI, which typically requires multiple breathold acquisitions. The air pockets, which are caught within the digestive pathway, induce spatially varying and large background field. T₂* map, that was calculated using UTE data, suffers from non-uniform T₂* value due to this background field, while does not appear in the local phase map of UTE data. Conclusion: Phase map generated using UTE mimicked the true field map even when non-zero susceptibility objects were present. Phase map generated by 3D GRE did not accurately mimic the true field map when non-zero susceptibility objects were present due to the significant field distortion as theoretically calculated. Nonetheless, UTE allows for phase maps to be free of susceptibility-induced distortion without the use of any post-processing protocols.

• The Transactions of the Korea Information Processing Society
• /
• v.2 no.2
• /
• pp.199-208
• /
• 1995

In this paper, a neural network architecture which can extract map symbols by being based on the results of physiological and neuropsychological studies on pattern recognition is proposed. This network is composed of multi-layers and synaptic activities of combining layers are implemented by spatial filters which approximate receptive fields of optic nerve cells. In pattern recognition which is followed by color classification for extracting of map symbols from input image, this network is searching for candidatepoints in lower layers (layer 2, 3) by using local features such as lines and end-points and then processing symbols recognition on those points in upper layer(layer 4) by using global features.

• Journal of the Institute of Convergence Signal Processing
• /
• v.10 no.1
• /
• pp.7-12
• /
• 2009

In this paper, we propose a directional feature extraction method for off-line handwritten numerals by using the morphological operations. Direction features are obtained from four directional line images, each of which contains horizontal, vertical, right-diagonal and left-diagonal lines in entire numeral lines. Conventional method for extracting directional features uses Kirsch masks which generate edge-shaped double line images for each direction, whereas our method uses directional erosion operations and generate single line images for each direction. To apply these directional erosion operations to the numeral image, preprocessing steps such as thinning and dilation are required, but resultant directional lines are more similar to numeral lines themselves. Our four [$4{\times}4$] directional features of a numeral are obtained from four directional line images through a zoning method. For obtaining the higher recognition rates of the handwrittern numerals, we use the multiple feature which is comprised of our proposed feature and the conventional features of a kirsch directional feature and a concavity feature. For recognition test with given features, we use a multi-layer perceptron neural network classifier which is trained with the back propagation algorithm. Through the experiments with the CENPARMI numeral database of Concordia University, we have achieved a recognition rate of 98.35%.

• The Transactions of the Korea Information Processing Society
• /
• v.6 no.2
• /
• pp.534-542
• /
• 1999

A sell-designed rate control strategy can improve overall picture quality for video transmission over a constant bit rate channel and the rate control method is not a normative part of MPEG-video standard, the performance of MPEG video codec can be quite different depends on how to implement the rate control scheme. The rate control scheme proposed in MPEG show good results when scene changes is not occurred. But it has weakness that it does not properly handle scene-changed pictures. Therefore picture quality after scene change is deteriorated, and possibility of overflow occurrence becomes high. In this paper, a new method for detection of scene change occurrence using local variance and a new determination scheme for adaptive quantization parameter, mqunt, which can consider local characteristic of an image by using previously computed the local variance from the scene change detection part are proposed. IN addition, and adaptive rate control scheme which can handles scene changed picture very efficiently by scene-changed picture is proposed. Computer simulations are performed to verify the performance of the proposed algorithm. The suggested detection algorithm precisely detected scene change. And the proposed rate control scheme shows better rate control performance as compared with that of the conventional MPEG scheme.

• Journal of KIISE:Software and Applications
• /
• v.36 no.12
• /
• pp.967-976
• /
• 2009

In this paper, we propose an automatic lower extremity vessel extraction based on rigid registration and bone elimination techniques in CT and CT angiography images. First, automatic partitioning of the lower extremity based on the anatomy is proposed to consider the local movement of the bone. Second, rigid registration based on distance map is performed to estimate the movement of the bone between CT and CT angiography images. Third, bone elimination and vessel masking techniques are proposed to remove bones in CT angiography image and to prevent the vessel near to bone from eroding. Fourth, post-processing based on vessel tracking is proposed to reduce the effect of misalignment and noises like a cartilage. For the evaluation of our method, we performed the visual inspection, accuracy measures and processing time. For visual inspection, the results of applying general subtraction, registered subtraction and proposed method are compared using volume rendering and maximum intensity projection. For accuracy evaluation, intensity distributions of CT angiography image, subtraction based method and proposed method are analyzed. Experimental result shows that bones are accurately eliminated and vessels are robustly extracted without the loss of other structure. The total processing time of thirteen patient datasets was 40 seconds on average.

• The Journal of The Korea Institute of Intelligent Transport Systems
• /
• v.12 no.6
• /
• pp.78-86
• /
• 2013

In this paper, we propose design of a high-performance vehicle imaging information system for an efficient vehicle imaging stabilization. The proposed system was designed the algorithm by divided as motion estimation and motion compensation. The motion estimation were configured as local motion vector estimation and irregular local motion vector detection, global motion vector estimation. The motion compensation was corrected for the four directions for compensate to the shake of vehicle video image using estimate GMV. The designed algorithm were designed the motion compensation technology chip by applied to IP for vehicle imaging stabilization. In this paper, the experimental results of the proposed vehicle imaging information system were proved to the effectiveness by compared with other methods, because imaging stabilization of moving vehicle was not used of memory by processing real-time. Also, it could be obtained to reduction effect of calculation time by arithmetic operation through to block matching.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.19 no.12
• /
• pp.2975-2980
• /
• 2015

Edge detection is one of image processing techniques applied for a variety of purposes in a number of areas and it is used as a necessary pretreatment process in most applications. Detect this edge has been conducted in various fields at domestic and international. In the conventional edge detection methods, there are Sobel, Prewitt, Roberts and LoG, etc using a fixed weights mask. Since conventional edge detection methods apply the images to the fixed weights mask, the edge detection characteristics appear somewhat insufficient. Therefore in this study, to complement this, preprocessing using gray-level transformation function and algorithm finding final edge using maximum and minimum value of estimated mask by local mask are proposed. And in order to assess the performance of proposed algorithm, it was compared with a conventional Sobel, Roberts, Prewitt and LoG edge detection methods.

• KIPS Transactions on Software and Data Engineering
• /
• v.10 no.12
• /
• pp.561-568
• /
• 2021

In recent years, autonomous driving technologies have become a high-value-added technology that attracts attention in the fields of science and industry. For smooth Self-driving, it is necessary to accurately detect an object and estimate its movement speed in real time. CNN-based deep learning algorithms and conventional dense optical flows have a large consumption time, making it difficult to detect objects and estimate its movement speed in real time. In this paper, using a single camera image, fast object detection was performed using the YOLOv5 algorithm, a deep learning algorithm, and fast estimation of the speed of the object was performed by using a local dense optical flow modified from the existing dense optical flow based on the detected object. Based on this algorithm, we present a system that can predict the collision time and probability, and through this system, we intend to contribute to prevent tram accidents.

• Journal of Electrical Engineering and Technology
• /
• v.13 no.6
• /
• pp.2468-2478
• /
• 2018

We propose the Sparse Feature Convolutional Neural Network (SFCNN) to reduce the volume of convolutional neural networks (CNNs). Despite the superior classification performance of CNNs, their enormous network volume requires high computational cost and long processing time, making real-time applications such as online-training difficult. We propose an advanced network that reduces the volume of conventional CNNs by producing a region-based sparse feature map. To produce the sparse feature map, two complementary region-based value extraction methods, cluster max extraction and local value extraction, are proposed. Cluster max is selected as the main function based on experimental results. To evaluate SFCNN, we conduct an experiment with two conventional CNNs. The network trains 59 times faster and tests 81 times faster than the VGG network, with a 1.2% loss of accuracy in multi-class classification using the Caltech101 dataset. In vehicle classification using the GTI Vehicle Image Database, the network trains 88 times faster and tests 94 times faster than the conventional CNNs, with a 0.1% loss of accuracy.