Earticle

An online surface inspection system m of hot rolled strips is introduced. This system is designed t o detect such main Surface defects on hot rolled strips as scar, scratches, pits, water drops Cracks. Cross hatchings, and so on. Multiple CCD area scan cameras are adopted to capture images of strip surface simultaneously, and all the images are processed by parallel computation system Real-time, which is supported by fast image process techniques and parallel computation techniques, in order to snap main defect regions on the surface of strips. At last, the defects will be classified to several types. The application of the system to practical production line shows that it can detect main defects of hot rolled strips more effectively than traditional method, and it is easily to be maintained.

Segmentation refers to a technique in which an image in digital form is partitioned into multiple segments (basically groups of pixels, also termed as Super pixels). This paper is a survey on Image Segmentation with its clustering techniques. Image Segmentation is the procedure of apportioning a picture into numerous segments, to change the exemplification of a picture into another which is more useful and easy to segment. A few universally useful calculations and approaches have been developed for picture division. It separates a digital picture into numerous locales to investigate them. It is likewise used to recognize segment items in the picture. A few picture segmentation procedures have been developed by the specialists with a specific end goal to make pictures smooth and simple to access. This paper describes segmentation techniques, advantages and disadvantages of the clustering methods and a comparison of the techniques.

The aim of this paper is to develop an efficient and a powerful Matlab based graphical user interface to address the problem of image segmentation. We propose two approaches for segmenting images: An automatic marker controlled watershed segmentation for segmenting an entire image or a scene and a semiautomatic graph cut based segmentation using fixation points. Automatic Watershed segmentation with a Sobel edge detector is used to detect the gradient of an input image resulting in an image less sensitive to noise. To deal with the usual problem of over segmentation using watershed, marker controlled watershed transformation is applied further for segmenting an image. Fixation based graph cut segmentation allows the user to analyze the input image displayed on the screen and specify some hard constraints indicating the object of interest or target object by using the mouse interaction. Experiments are done on the publically available dataset and the results of the supervised evaluation methods are observed to be satisfactory and are demonstrated along with the manually segmented reference image or a ground truth image obtained from segmentation evaluation database

The importance of this paper is to give a survey of advanced picture division methods. The issues of digital picture division speak to formidable difficulties for PC vision. The extensive variety of the issues of PC vision may make excellent utilisation of picture division. This paper examines and assesses the diverse strategies for division systems. We examine the fundamental propensity of every calculation with their applications, points of interest and burdens. This examination is helpful for deciding the appropriate utilisation of the picture division strategies and for enhancing their exactness and execution and furthermore for the primary goal, which planning new calculations.

본 논문에서는 TFT-LCD에 사용되는 편광 필름(polarized film)의 제작 과정 중 최종 단계에서 수행되는 필름의 결함 검출 및 결함 유형을 판정하기 위한 필름 결함 검출 시스템(Film Defect Inspection System: FDIS)을 설계하고 이를 구현하였다. 제안한 시스템은 영상 세그멘테이션 기법을 이용하여 편광 필름 영상으로부터 결함을 검출하였고, 검출된 결함의 영상을 분석하여 결함 유형을 판정할 수 있도록 설계되었다. 결함 유형의 판정은 결함 영역의 형태적 특성 및 질감(texture) 등의 특징을 추출하여 템플리트(template) 데이타베이스에 저장된 기준(reference) 결함 영상과 비교함으로써 수행된다. FDIS를 이용한 실험 결과, 테스트 영상에서 모든 결함 영역을 빠른 시간 안에 (평균 0.64초), 정확히 검출하였으며(Precision 1.0, Recall 1.0), 결함 유형을 판정하는 실험에서도 평균 Precision 0.96, Recall 0.95로 정확도가 매우 높은 것을 관찰할 수 있었다. 또한 회전 변형을 적용한 경우의 결함 유형 검출 실험에서도 평균 Precision 0.95, Recall 0.89로 제안한 기법이 회전 변환에 대하여 견고함을 보여 주었다.

In this paper, we design and implement the Film Defect Inspection System (FDIS) that detects film defects and determines their types which can be used for producing polarized films of TFT-LCD. The proposed system is designed to detect film defects from polarized film images using image segmentation techniques and to determine defect types through the image analysis of detected defects. To determine defect types, we extract features such as shape and texture of defects, and compare those features with corresponding features of referential images stored in a template database. Experimental results using FDIS show that the proposed system detects all defects of test images effectively (Precision 1.0, Recall 1.0) and efficiently (within 0.64 second in average), and achieves the considerably high correctness in determining defect types (Precision 0.96 and Recall 0.95 in average). In addition, our system shows the high robustness for rotated transformation of images, achieving Precision 0.95 and Recall 0.89 in average.

In Korean medicine, tongue diagnosis is one of the important diagnostic methods for diagnosing abnormalities in the body. Representative features that are used in the tongue diagnosis include color, shape, texture, cracks, and tooth marks. When diagnosing a patient through these features, the diagnosis criteria may be different for each oriental medical doctor, and even the same person may have different diagnosis results depending on time and work environment. In order to overcome this problem, recent studies to automate and standardize tongue diagnosis using machine learning are continuing and the basic process of such a machine learning-based tongue diagnosis system is tongue segmentation. In this paper, image data is augmented based on the main tongue features, and backbones of various famous deep learning architecture models are used for automatic tongue segmentation. The experimental results show that the proposed augmentation technique improves the accuracy of tongue segmentation, and that automatic tongue segmentation can be performed with a high accuracy of 99.12%.

항공우주 산업의 발전으로 인해 인공지능을 활용하여 인공위성 영상에서 관심 객체를 분석하는 연구가 증가하고 있다. 그러나 위성 영상은 일반적인 8bit RGB 카메라 영상과 달리 16bit 픽셀 값을 주로 가지며, 이를 그대로 사용하면 이상치 값으로 인해 영상이 어두워지는 문제가 발생할 수 있다. 이러한 문제는 객체 식별을 어렵게 만들고 분석 성능에 부정적인 영향을 미친다. 이 문제를 해결하기 위해 다양한 영상 향상 기법이 제안되었으나, 위성마다 세부 특성이 다르고 수행하고자 하는 작업에 따라 각 기법의 효과가 달라지므로 적합한 기법을 검토할 필요가 있다. 본 연구에서는 국내에서 널리 활용되는 KOMPSAT-3A 위성의 데이터셋을 사용해, 5개의 영상 향상 기법 중 semantic segmentation 작업에 적합한 영상 향상 기법에 대해 분석하였다. 5가지 semantic segmentation 모델을 사용한 실험 결과, 백분위수 스트레칭이 3가지 모델에서 좋은 성능을 보여 가장 보편적으로 활용 가능한 방법으로 판단되었다. 또한, 도시 분석에서 중요한 객체인 건물과 도로의 경우, recursive separated and weighted histogram equalization (RSWHE)과 백분위수 스트레칭이 효과적인 것으로 나타났다.

Advances in the aerospace industry have driven growing research into the use of artificial intelligence for analyzing objects of interest in satellite imagery. Unlike typical 8-bit RGB camera images, however, satellite imagery often contains 16-bit pixel values, which can result in outliers that darken the image. This issue leads to difficulty in object identification and negatively impacts analysis performance. To address this issue, various image enhancement techniques have been proposed, but the effectiveness of each technique depends on the specifics of each satellite and the task to be performed. To address this issue, various image enhancement techniques have been proposed. However, since each satellite has unique characteristics and the effectiveness of each technique varies depending on the specific task, it is necessary to carefully evaluate which technique is most suitable. This research analyzed which of the five image enhancement techniques is most suitable for semantic segmentation tasks using the dataset from the KOMPSAT-3A satellite, which is widely used in South Korea. Experimental results using five semantic segmentation models indicated that percentile stretching performed well in three models, suggesting it as the most universally applicable method. In addition, for buildings and roads, which are important objects in urban analysis, recursive separated and weighted histogram equalization (RSWHE) and percentile stretching were found to be effective.