• Journal of Platform Technology
• /
• 제8권1호
• /
• pp.16-23
• /
• 2020

상표권 보호를 위한 상표 이미지의 진위 결정에 심층 신경망인 합성곱 신경망을 이용할 수 있다. 이를 위해, 상표로 등록되어 있는 한 장의 상표 이미지를 반복적으로 학습하는 것은 기계학습의 성능을 감소시키는 원인이 된다. 따라서, 이러한 응용에서 학습 데이터는 다양한 방법으로 생성된다. 하지만 대상 이미지가 회전되어 있으면 원본이라 하더라도 인식하지 못하거나 위조 상표로 분류되기도 한다. 본 논문에서는 회전한 상표 이미지의 진위 결정을 위한 기계학습 데이터의 확장 방법을 제안한다. 본 논문에서 제안하는 학습 데이터 확장 방법은 기울어진 이미지를 생성하고 이를 학습 데이터로 사용하는 것이다. 본 논문에서 제안하는 학습 데이터 확장 방법의 유효성 검증을 위해 대학의 로고를 대상으로 학습 데이터를 생성하였으며, 이를 활용하여 합성곱 신경망을 학습시킨 후 검증용 데이터를 이용하여 정확도를 평가하였다. 정확도 평가 결과에 따르면 본 논문에서 제안한 방법으로 생성한 학습 데이터를 활용하면 회전한 상표를 대상으로 한 진위 여부 결정에 합성곱 신경망을 활용할 수 있다.

• 한국컴퓨터정보학회논문지
• /
• 제23권4호
• /
• pp.147-153
• /
• 2018

In this paper, we propose an efficient WBC 14-Diff classification which performs using the WBC-ResNet-152, a type of CNN model. The main point of view is to use Super-pixel for the segmentation of the image of WBC, and to use ResNet for the classification of WBC. A total of 136,164 blood image samples (224x224) were grouped for image segmentation, training, training verification, and final test performance analysis. Image segmentation using super-pixels have different number of images for each classes, so weighted average was applied and therefore image segmentation error was low at 7.23%. Using the training data-set for training 50 times, and using soft-max classifier, TPR average of 80.3% for the training set of 8,827 images was achieved. Based on this, using verification data-set of 21,437 images, 14-Diff classification TPR average of normal WBCs were at 93.4% and TPR average of abnormal WBCs were at 83.3%. The result and methodology of this research demonstrates the usefulness of artificial intelligence technology in the blood cell image classification field. WBC-ResNet-152 based morphology approach is shown to be meaningful and worthwhile method. And based on stored medical data, in-depth diagnosis and early detection of curable diseases is expected to improve the quality of treatment.

• 대한원격탐사학회지
• /
• 제26권3호
• /
• pp.317-324
• /
• 2010

In recent years, many automatic classification approaches have been employed. An automatic classification method can be effective, time-saving and can produce objective results due to the exclusion of operator intervention. This paper proposes a classification method based on automated training for high resolution multispectral images using ancillary data. Generally, it is problematic to automatically classify high resolution images using ancillary data, because of the scale difference between the high resolution image and the ancillary data. In order to overcome this problem, the proposed method utilizes the classification results of a Landsat image as a medium for automatic classification. For the classification of a Landsat image, a maximum likelihood classification is applied to the image, and the attributes of ancillary data are entered as the training data. In the case of a high resolution image, a K-means clustering algorithm, an unsupervised classification, was conducted and the result was compared to the classification results of the Landsat image. Subsequently, the training data of the high resolution image was automatically extracted using regular rules based on a RELATIONAL matrix that shows the relation between the two results. Finally, a high resolution image was classified and updated using the extracted training data. The proposed method was applied to QuickBird and SPOT-5 images of non-accessible areas. The result showed good performance in accuracy assessments. Therefore, we expect that the method can be effectively used to automatically construct thematic maps for non-accessible areas and update areas that do not have any attributes in geographic information system.

• 한국방송∙미디어공학회:학술대회논문집
• /
• 한국방송공학회 2009년도 IWAIT
• /
• pp.355-359
• /
• 2009

In this paper, we describe self-training super-resolution. Our approach is based on example based algorithms. Example based algorithms need training images, and selection of those changes the result of the algorithm. Consequently it is important to choose training images. We propose self-training based super-resolution algorithm which use an input image itself as a training image. It seems like other example based super-resolution methods, but we consider training phase as the step to collect primitive information of the input image. And some artifacts along the edge are visible in applying example based algorithms. We reduce those artifacts giving weights in consideration of the edge direction. We demonstrate the performance of our approach is reasonable several synthetic images and real images.

• 로봇학회논문지
• /
• 제18권3호
• /
• pp.293-298
• /
• 2023

The military is facing a continuous decrease in personnel, and in order to cope with potential accidents and challenges in operations, efforts are being made to reduce the direct involvement of personnel by utilizing the latest technologies. Recently, the use of various sensors related to Manned-Unmanned Teaming and artificial intelligence technologies has gained attention, emphasizing the need for flexible utilization methods. In this paper, we propose four dataset construction methods that can be used for effective training of robots that can be deployed in military operations, utilizing not only RGB image data but also data acquired from IR image sensors. Since there is no publicly available dataset that combines RGB and IR image data, we directly acquired the dataset within buildings. The input values were constructed by combining RGB and IR image sensor data, taking into account the field of view, resolution, and channel values of both sensors. We compared the proposed method with conventional RGB image data classification training using the same learning model. By employing the proposed image data fusion method, we observed improved stability in training loss and approximately 3% higher accuracy.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제12권7호
• /
• pp.3172-3193
• /
• 2018

Recently, several studies have shown that linear representation based approaches are very effective and efficient for image classification. One of these linear-representation-based approaches is the Collaborative representation (CR) method. The existing algorithms based on CR have two major problems that degrade their classification performance. First problem arises due to the limited number of available training samples. The large variations, caused by illumintion and expression changes, among query and training samples leads to poor classification performance. Second problem occurs when an image is partially noised (contiguous occlusion), as some part of the given image become corrupt the classification performance also degrades. We aim to extend the collaborative representation framework under limited training samples face recognition problem. Our proposed solution will generate virtual samples and intra-class variations from training data to model the variations effectively between query and training samples. For robust classification, the image patches have been utilized to compute representation to address partial occlusion as it leads to more accurate classification results. The proposed method computes representation based on local regions in the images as opposed to CR, which computes representation based on global solution involving entire images. Furthermore, the proposed solution also integrates the locality structure into CR, using Euclidian distance between the query and training samples. Intuitively, if the query sample can be represented by selecting its nearest neighbours, lie on a same linear subspace then the resulting representation will be more discriminate and accurately classify the query sample. Hence our proposed framework model the limited sample face recognition problem into sufficient training samples problem using virtual samples and intra-class variations, generated from training samples that will result in improved classification accuracy as evident from experimental results. Moreover, it compute representation based on local image patches for robust classification and is expected to greatly increase the classification performance for face recognition task.

• 융합신호처리학회 학술대회논문집
• /
• 한국신호처리시스템학회 2003년도 하계학술대회 논문집
• /
• pp.94-97
• /
• 2003

본 논문에서는, 웨이브렛 변환과 잡음 섞인 숫자 영상에 대한 최적화 인식 훈련기법을 사용한 다계층 신경망을 제안하고, 이 시스템을 아라비아숫자 인식에 적용한다. 웨이브렛 변환을 이용해 원 영상 정보의 중요한 부분은 최대한 보존하면서 입력벡터의 크기를 줄임으로써 신경망의 노드 수와 학습 수렴시간이 줄어들도록 하였고, 최적화 인식 훈련기법은 데이터의 잡음을 점차적으로 높여가면서 훈련벡터에 적용, 인식률의 변화에 대해 살펴보았다. 잡음이 섞인 숫자 영상의 인식율을 높이기 위해 원 영상에 0, 10, 20, 30, 40, 50㏈의 잡음을 섞은 영상을 훈련에 함께 사용하였다. 테스트 영상에 잡음이 30∼50㏈정도 섞였을 경우에는 원 영상만을 훈련에 이용했을 패와 잡음이 섞인 영상을 이용하여 훈련시켰을 경우에 인식율의 차이가 별로 없지만, 0∼20㏈정도 섞인 영상을 테스트에 사용할때에는 0, 10, 20, 30, 40 , 50㏈의 잡음이 있는 영상을 훈련에 사용했을 때가 원 영상만을 훈련에 이용했을 경우에 비해 인식율이 9% 향상된다.

• 천문학회보
• /
• 제44권2호
• /
• pp.78.1-78.1
• /
• 2019

There are a huge number of faint objects that have not been observed due to the lack of large and deep surveys. In this study, we demonstrate that a deep learning approach can produce a better quality deep image from a single pass imaging so that could be an alternative of conventional image stacking technique or the expensive large and deep surveys. Using data from the Sloan Digital Sky Survey (SDSS) stripe 82 which provide repeatedly scanned imaging data, a training data set is constructed: g-, r-, and i-band images of single pass data as an input and r-band co-added image as a target. Out of 151 SDSS fields that have been repeatedly scanned 34 times, 120 fields were used for training and 31 fields for validation. The size of a frame selected for the training is 1k by 1k pixel scale. To avoid possible problems caused by the small number of training sets, frames are randomly selected within that field each iteration of training. Every 5000 iterations of training, the performance were evaluated with RMSE, peak signal-to-noise ratio which is given on logarithmic scale, structural symmetry index (SSIM) and difference in SSIM. We continued the training until a GAN model with the best performance is found. We apply the best GAN-model to NGC0941 located in SDSS stripe 82. By comparing the radial surface brightness and photometry error of images, we found the possibility that this technique could generate a deep image with statistics close to the stacked image from a single-pass image.

• 반도체디스플레이기술학회지
• /
• 제17권1호
• /
• pp.71-75
• /
• 2018

Image enhancement is a necessary end essential step after taking a picture with a digital camera. Many different photo software packages attempt to automate this process with various auto enhancement techniques. This paper provides and implements a system that can learn a user's preferences and apply the preferences into the process of image enhancement. Five major components are applied to the implemented system, which are computing a distance metric, finding a training set, finding an optimal parameter set, training and finally enhancing the input image. To estimate the validity of the method, we carried out user studies, and the fact that the implemented system was preferred over the method without learning user preferences.

• Design & Manufacturing
• /
• 제15권3호
• /
• pp.1-6
• /
• 2021

An artificial neural network model based on a deep learning algorithm is known to be more accurate than humans in image classification, but there is still a limit in the sense that there needs to be a lot of training data that can be called big data. Therefore, various techniques are being studied to build an artificial neural network model with high precision, even with small data. The transfer learning technique is assessed as an excellent alternative. As a result, the purpose of this study is to develop an artificial neural network system that can classify burr images of light guide plate products with 99% accuracy using transfer learning technique. Specifically, for the light guide plate product, 150 images of the normal product and the burr were taken at various angles, heights, positions, etc., respectively. Then, after the preprocessing of images such as thresholding and image augmentation, for a total of 3,300 images were generated. 2,970 images were separated for training, while the remaining 330 images were separated for model accuracy testing. For the transfer learning, a base model was developed using the NASNet-Large model that pre-trained 14 million ImageNet data. According to the final model accuracy test, the 99% accuracy in the image classification for training and test images was confirmed. Consequently, based on the results of this study, it is expected to help develop an integrated AI production management system by training not only the burr but also various defective images.