• Journal of KIISE:Computing Practices and Letters
• /
• v.16 no.6
• /
• pp.654-662
• /
• 2010

In this paper, we suggest an advanced algorithm, to recognize pedestrian/non-pedestrian using second-stage cascade method, which applies Adaboost algorithm to make a strong classification from weak classifications. First, we extract two feature vectors: (i) Histogram of Oriented Gradient (HOG) which includes gradient information and differential magnitude; (ii) Curvature-HOG which is based on four different curvature features per pixel. And then, a strong classification needs to be obtained from weak classifications for composite recognition method using both HOG and curvature-HOG. In the proposed method, we use one feature vector and one strong classification for the first stage of recognition. For the recognition-failed image, the other feature and strong classification will be used for the second stage of recognition. Based on our experiment, the proposed algorithm shows higher recognition rate compared to the traditional method.

• Journal of Internet of Things and Convergence
• /
• v.6 no.1
• /
• pp.83-95
• /
• 2020

Being a dense and complex task of precisely detecting the weeds in practical crop field environment, previous approaches lack in terms of speed of processing image frames with accuracy. Although much of the attention has been given to classify the plants diseases but detecting crop weed issue remained in limelight. Previous approaches report to use fast algorithms but inference time is not even closer to real time, making them impractical solutions to be used in uncontrolled conditions. Therefore, we propose a detection model for the complex rice weed detection task. Experimental results show that inference time in our approach is reduced with a significant margin in weed detection task, making it practically deployable application in real conditions. The samples are collected at two different growth stages of rice and annotated manually

• The Korean Journal of Pesticide Science
• /
• v.15 no.3
• /
• pp.269-277
• /
• 2011

Recent outbreak of new diseases and pests which were introduced from abroad, seriously hampered both quality and safety of paprika fruits. This study has been carried out to aid an establishment of guideline for safe use of pesticides and reduction of their residues on paprika. Systemic fungicides boscalid and pyraclostrobin of either mixed (a.i.; 13.6+6.8%) or single (a.i.; 47 and 18.8%, respectively) water dispersible granule formulation(WG) products were sprayed with recommended or double dosage on paprika grown in green house at March and June. To draw pre-harvest residue limit, residues of each fungicide were analyzed from fruits collected eight times from 18 to 1 day pre-harvest. The biological half-lives of both boscalid and pyraclostrobin in mixed formulation in March and June were slightly shorter than those of single formulation which ranged from 14.4 to 20.1 days. Residue levels of both fungicides of single formulation in fruits in June were about one lower compared to those in March. However, application of double dosage frequently exceeded MRLs from fruits grown both seasons. These results showed that residue levels on fruits persisted longer period of time, more than two weeks, and so the case applied in winter season. The dissipation of fungicides on leaves and fruits was compared. The distribution of both fungicides in leaves was 20-200 times higher than that of fruits and persisted up to 18 days of pre-harvest period at the concentration of 10-40 ${\mu}g\;g^{-1}$. This study indicated that the mixed formulation product exhibited low residues in fruits, but high and long enough to pathogen growth in leaves.

• Proceedings of the Korea Institute of Convergence Signal Processing
• /
• 2001.06a
• /
• pp.165-168
• /
• 2001

본 논문에서는 카메라 또는 카메라 플랫폼의 흔들림 등 외부 영향과 비디오 시퀀스내의 모션이 함께 존재할 경우 출렁이는 비디오를 전자적으로 안정화하는 방법을 제안한다. 일반적인 영상 안정 시스템은 모션 측정과 모션 보상의 두 과정으로 구성되는데 모션 측정에서는 프레임간 모션 모델을 가정하고 파라메타를 측정하며 모션 보상에서는 측정된 파라메타를 이용하여 모션을 보상한다. 영상 내에 카메라 모션 이외의 움직임이 있을 경우 파라메타의 측정을 일관성 없게 만들 수 있으므로 이를 해결하기 위해 MVSD(Motion Vector Scatter Diagram)에 기반한 영상 안정 방법이 제안되었다. 그러나 이 방법은 최적화 파라메타를 정량화 하는데 한계가 있고 또한 계산 시간이 오래 걸리는 단점이 있어 이의 해결을 위해 본 논문에서는 선 기반(Line-based) 영상 안정 방법을 제안한다. 이 방법은 먼저 기준 영상에서 median filter를 이용해 영상 내의 코너를 검출하고 특징적인 두 점을 선택하여 이를 선으로 연결한다. 현재 영상에서 correlation을 이용하여 상응하는 두 특징점을 찾고 subpixel 방법으로 정확한 위치를 계산하여 선을 구한다. 이 두 선을 일치시키는 과정에서 모션 파라메타를 구하는데 먼저 평행 이동을 통해 한쪽 글을 일치시키고 이 과정에서 translation x, y 파라메타를 구한다. 다음 단계에서 한 쪽 끝이 일치된 두 선이 이루는 각을 계산하여 rotation 파라메타를 구한다. 이 방법으로 구해진 파라메타를 이용하여 모션 보상을 함으로서 영상 안정을 이를 수 있었다.

• Journal of Broadcast Engineering
• /
• v.27 no.6
• /
• pp.923-935
• /
• 2022

This paper deals with detection and classification of leaf diseases for phenomics systems. As the smart farm systems of plants are increased, It is important to determine quickly the abnormal growth of plants without supervisors. This paper considers the color distribution and shape information of leaf diseases, and designs two deep leaning networks in training the leaf diseases. In the first step, color distribution of input image is analyzed for possible diseases. In the second step, the image is first partitioned into small segments using mean shift clustering, and the color information of each segment is inspected by the proposed Color Network. When a segment is determined as disease, the shape parameters of the segment are extracted and inspected by proposed Shape Network to classify the leaf disease types in the third step. According to the experiments with two types of diseases (frogeye/rust and tipburn) for apple leaves and iceberg, the leaf diseases are detected with 92.3% recall for a segment and with 99.3% recall for an input image where there are usually more than two disease segments. The proposed method is useful for detecting leaf diseases quickly in the smart farm environment, and is extendible to various types of new plants and leaf diseases without additional learning.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.29 no.3C
• /
• pp.374-386
• /
• 2004

The mammogram provides the way to observe detailed internal organization of breasts to radiologists for the early detection. This paper is mainly focused on efficiently detecting the Microcalcification's Region Of Interest(ROI)s. Breast cancers can be caused from either microcalcifications or masses. Microcalcifications are appeared in a digital mammogram as tiny dots that have a little higher gray levels than their surrounding pixels. We can roughly determine the area which possibly contain microcalifications. In general, it is very challenging to find all the microcalcifications in a digital mammogram, because they are similar to some tissue parts of a breast. To efficiently detect microcalcifications ROI, we used four sequential processes; preprocessing for breast area detection, modified multilevel thresholding, ROI selection using simple thresholding filters and final ROI selection with two stages of neural networks. The filtering process with boundary conditions removes easily-distinguishable tissues while keeping all microcalcifications so that it cleans the thresholded mammogram images and speeds up the later processing by the average of 86％. The first neural network shows the average of 96.66％ recognition rate. The second neural network performs better by showing the average recognition rate 98.26％. By removing all tissues while keeping microcalcifications as much as possible, the next parts of a CAD system for detecting breast cancers can become much simpler.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.16 no.3
• /
• pp.2185-2191
• /
• 2015

With the advance of the environment where we can get various multimedia contents, adult image detection has become an important issue these days. In this paper, we suggest a method of robustly detecting navel areas from input images which can be usefully utilized in adult image detection. The suggested algorithm first extracts face regions and extracts candidate nipple areas using a nipple map. Our method then selects only actual nipple regions by filtering candidate areas with geometrical features and an average nipple filter. Subsequently, the method robustly detects navel areas by using the structural relation with the nipple areas and applying edge and saturation images. Experimental results show that the suggested algorithm can effectively detect navel regions.

• KIPS Transactions on Software and Data Engineering
• /
• v.3 no.9
• /
• pp.361-368
• /
• 2014

The LPR(License plate recognition) system has been developed to efficient control for complex traffic environment and currently be used in many places. However, because of light, noise, background changes, environmental changes, damaged plate, it only works limited environment, so it is difficult to use in real-time. This paper presents a heuristic segmentation algorithm for robust to noise and illumination changes and introduce a real-time license plate recognition system using it. In first step, We detect the plate utilized Haar-like feature and Adaboost. This method is possible to rapid detection used integral image and cascade structure. Second step, we determine the type of license plate with adaptive histogram equalization, bilateral filtering for denoise and segment accurate character based on adaptive threshold, pixel projection and associated with the prior knowledge. The last step is character recognition that used histogram of oriented gradients (HOG) and multi-layer perceptron(MLP) for number recognition and support vector machine(SVM) for number and Korean character classifier respectively. The experimental results show license plate detection rate of 94.29%, license plate false alarm rate of 2.94%. In character segmentation method, character hit rate is 97.23% and character false alarm rate is 1.37%. And in character recognition, the average character recognition rate is 98.38%. Total average running time in our proposed method is 140ms. It is possible to be real-time system with efficiency and robustness.

• Journal of the Korea Society of Computer and Information
• /
• v.26 no.1
• /
• pp.45-55
• /
• 2021

This paper has proposed an algorithm that detecting for dense small vehicle in large image efficiently. It is consisted of two Ensemble Deep-Learning Network algorithms based on Coarse to Fine method. The system can detect vehicle exactly on selected sub image. In the Coarse step, it can make Voting Space using the result of various Deep-Learning Network individually. To select sub-region, it makes Voting Map by to combine each Voting Space. In the Fine step, the sub-region selected in the Coarse step is transferred to final Deep-Learning Network. The sub-region can be defined by using dynamic windows. In this paper, pre-defined mapping table has used to define dynamic windows for perspective road image. Identity judgment of vehicle moving on each sub-region is determined by closest center point of bottom of the detected vehicle's box information. And it is tracked by vehicle's box information on the continuous images. The proposed algorithm has evaluated for performance of detection and cost in real time using day and night images captured by CCTV on the road.

• Journal of Korea Multimedia Society
• /
• v.12 no.7
• /
• pp.929-940
• /
• 2009

This paper presents multipurpose watermarking scheme in coutourlet transform domain for copyright protection, authentication and transform detection. Since contourlet transform can detect more multi direction edge and smooth contour than wavelet transform, the proposed scheme embeds multi watermarks in contourlet domain based on 4-level Laplacian pyramid and 2-level directional filter bank. In the first stage of the robust watermarking scheme for copyright protection, we generates the sequence of circle patterns according to watermark bits and projects these patterns into the average of magnitude coefficients of high frequency directional subbands. Then the watermark bit is embedded into variance distribution of the projected magnitude coefficients. In the second stage that is the semi-fragile watermarking scheme for authentication and transform detection, we embed the binary watermark image in the low frequency subband of higher level by using adaptive quantization modulation scheme. From the evaluation experiment using Checkmark 2.1, we verified that the proposed scheme is superior to the conventional scheme in a view of the robustness and the invisibility.