• Journal of the Institute of Electronics and Information Engineers
• /
• v.50 no.5
• /
• pp.287-294
• /
• 2013

Binary connected-component labeling is widely used in the fields of the image processing and the computer vision. Many kinds of labeling techniques have been developed, and two-scan is known as the fastest method among them. Traditionally pixel-based scan masks have been used for the first stage of the two-scan. Recently, block-based labeling techniques were introduced by C. Grana et. al. and L. He et. al. They are faster than pixel-based labeling methods. In this paper, we propose a new binary connected-component labeling technique with block-based labels and a pixel-based scan mask. The experimental results with various images show that the proposed method is faster than the He's which is known as the fastest method currently. The amount of performance enhancement is averagely from 3.9% to 22.4% according to the sort of the images.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2007.06a
• /
• pp.721-723
• /
• 2007

Recently, the size and location of the acquired container image while the container is loading and unloading in Harbors is not fixed. And it is difficult to get a good image for recognition because of the variation of external environment as those the size of container and where the yard-tractor stop is. In this paper, we estimate where the container ISO-code set is using Top-hat transform from realtime images and get an image to recognize container ISO-code using PAN/TILT/ZOOM camera. We extract the container ISO-code using Top-hat transform and Histogram projection. After binarization, we extract each character from complex background using labeling. We use BP(Backpropagation Network) to recognize extracted characters.

• Fisheries and Aquatic Sciences
• /
• v.4 no.1
• /
• pp.10-17
• /
• 2001

One serious problem among the troubles to identify objects in an optical microscopic image is contour background due to non-uniform light source and various transparency of samples. To solve this problem, this paper proposed an elimination method of the contour background and compensation technique as follows. First, Otsu's optimal thresholding method extracts pixels representing background. Second, bilinear interpolation finds non-deterministic background pixels among the sampled pixels. Third, the 2D cubic fitting method composes surface function from pivoted background pixels. Fourth, reconstruction procedure makes a contour image from the surface function. Finally, elimination procedure subtracts the approximated background from the original image. To prove the effectiveness of the proposed algorithm, this algorithm is applied to the yeast Zygosaccharomyces rouxii and ammonia-oxidizing bacteria Acinetobacter sp. Labeling by this proposed method can remove some noise and is more exact than labeling by only Otsu's method. Futhermore, we show that it is more effective for the reduction of noise.

• Journal of Institute of Control, Robotics and Systems
• /
• v.21 no.9
• /
• pp.858-862
• /
• 2015

This paper presents a new forward-viewing mono-camera based obstacle detection algorithm for mobile robots. The proposed method extracts the coarse location of an obstacle in an image using inverse perspective mapping technique from sequential images. In the next step, graph-cut based image labeling is conducted for estimating the exact obstacle boundary. The graph-cut based labeling algorithm labels the image pixels as either obstacle or floor as the final outcome. Experiments are performed to verify the obstacle detection performance of the developed algorithm in several examples, including a book, box, towel, and flower pot. The low illumination condition, low color contrast between floor and obstacle, and floor pattern cases are also tested.

• Journal of IKEEE
• /
• v.26 no.4
• /
• pp.647-652
• /
• 2022

In this paper, we propose a prediction system for skin pore labeling based on a CNN(Convolution Neural Network) model, where a data set is constructed by processing skin images taken by users, and a pore feature image is generated by the proposed image processing algorithm. The skin image data set was labeled for pore characteristics based on the visual classification criteria of skin beauty experts. The proposed image processing algorithm was applied to generate pore feature images from skin images and to train a CNN model that predicts pore feature ratings. The prediction results with pore features by the proposed CNN model is similar to experts visual classification results, where less learning time and higher prediction results were obtained than the results by the comparison model (Resnet-50). In this paper, we describe the proposed image processing algorithm and CNN model, the results of the prediction system and future research plans.

• Journal of the Korea Computer Graphics Society
• /
• v.28 no.3
• /
• pp.55-65
• /
• 2022

Recently, deep learning-based automated systems for identifying and detecting landmarks have been proposed. In order to train such a deep learning-based model without overfitting, a large amount of image and labeling data is required. Conventionally, an experienced reader manually identifies and labels landmarks in a patient's image. However, such measurement is not only expensive, but also has poor reproducibility, so the need for an automated labeling method has been raised. In addition, in the X-ray image, since various human tissues on the path through which the photons pass are displayed, it is difficult to identify the landmark compared to a general natural image or a 3D image modality image. In this study, we propose a geometric data augmentation technique that enables the generation of a large amount of labeling data in X-ray images. In addition, the optimal attention mechanism for landmark detection was presented through the implementation and application of various attention techniques to improve the detection performance of 16 major landmarks in the skull. Finally, among the major cranial landmarks, markers that ensure stable detection are derived, and these markers are expected to have high clinical application potential.

• Journal of Advanced Navigation Technology
• /
• v.19 no.6
• /
• pp.484-491
• /
• 2015

Aerial scene is greatly increased by the introduction and supply of the image due to the growth of digital optical imaging technology and development of the UAV. It has been used as the extraction of ground properties, classification, change detection, image fusion and mapping based on the aerial image. In particular, in the image analysis and utilization of deep learning algorithm it has shown a new paradigm to overcome the limitation of the field of pattern recognition. This paper presents the possibility to apply a more wide range and various fields through the segmentation and classification of aerial scene based on the Deep learning(ConvNet). We build 4-classes image database consists of Road, Building, Yard, Forest total 3000. Each of the classes has a certain pattern, the results with feature vector map come out differently. Our system consists of feature extraction, classification and training. Feature extraction is built up of two layers based on ConvNet. And then, it is classified by using the Multilayer perceptron and Logistic regression, the algorithm as a classification process.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.17 no.4
• /
• pp.935-944
• /
• 2013

In this paper, a new illumination compensation algorithm by segmentation with depth information is proposed to improve the coding efficiency of multi-view images. In the proposed algorithm, a reference image is first segmented into several layers where each layer is composed of objects with a similar depth value. Then we separate objects from each other even in the same layer by labeling each separate region in the layered image. Then, the labeled reference depth image is converted to the position of the distortion image view by using 3D warping algorithm. Finally, we apply an illumination compensation algorithm to each of matched regions in the converted reference view and distorted view. The occlusion regions that occur by 3D warping are also compensated by a global compensation method. Through experimental results, we are able to confirm that the proposed algorithm has better performance to improve coding efficiency.

• Journal of the Institute of Electronics and Information Engineers
• /
• v.50 no.11
• /
• pp.188-194
• /
• 2013

In this paper, we proposed a new method for pulmonary vessels image segmentation and refinement from pulmonary image. Proposed method consist of following five steps. First, threshold estimation is performed by polynomial regression analysis of histogram variation rate of the pulmonary image. Second, segmentation of pulmonary vessels object is performed by density-based segmentation method based on estimated threshold in first step. Third, 2D connected component labeling method is applied to segmented pulmonary vessels. The seed point of both side diaphragms is determined by eccentricity and size of component. Fourth step is diaphragm extraction by 3D region growing method at the determined seed point. Finally, noise cancelation of pulmonary vessels image is performed by 3D connected component labeling method. The experimental result is showed accurately pulmonary vessels image segmentation, the diaphragm extraction and the noise cancelation of the pulmonary vessels image.

• Journal of Korean Tunnelling and Underground Space Association
• /
• v.21 no.3
• /
• pp.419-432
• /
• 2019

Most of deep learning model training was proceeded by supervised learning, which is to train labeling data composed by inputs and corresponding outputs. Labeling data was directly generated manually, so labeling accuracy of data is relatively high. However, it requires heavy efforts in securing data because of cost and time. Additionally, the main goal of supervised learning is to improve detection performance for 'True Positive' data but not to reduce occurrence of 'False Positive' data. In this paper, the occurrence of unpredictable 'False Positive' appears by trained modes with labeling data and 'True Positive' data in monitoring of deep learning-based CCTV accident detection system, which is under operation at a tunnel monitoring center. Those types of 'False Positive' to 'fire' or 'person' objects were frequently taking place for lights of working vehicle, reflecting sunlight at tunnel entrance, long black feature which occurs to the part of lane or car, etc. To solve this problem, a deep learning model was developed by simultaneously training the 'False Positive' data generated in the field and the labeling data. As a result, in comparison with the model that was trained only by the existing labeling data, the re-inference performance with respect to the labeling data was improved. In addition, re-inference of the 'False Positive' data shows that the number of 'False Positive' for the persons were more reduced in case of training model including many 'False Positive' data. By training of the 'False Positive' data, the capability of field application of the deep learning model was improved automatically.