• Journal of Korean Society for Geospatial Information Science
• /
• v.24 no.1
• /
• pp.121-128
• /
• 2016

Many studies are focused on image data and classifier for comparison or improvement of classification accuracy. Therefore studies are needed aspect of the training samples on supervised classification which depend on reference data or skill of analyst. This study tries to assess usability of field spectra as training samples on supervised classification. Classification accuracies of hyperspectral and multispectral images were assessed using training samples from image itself and field spectra, respectively. The results shown about 90% accuracy with training sample collected from image. Using field spectra as training sample, accuracy was decreased 10%p for hyperspectral image, and 20%p for multispectral image. Especially, some classes shown very low accuracies due to similar spectral characteristics on multispectral image. Therefore, field spectra might be used as training samples on classification of hyperspectral image, although it has limitation for multispectral image.

• Korean Journal of Remote Sensing
• /
• v.2 no.1
• /
• pp.13-21
• /
• 1986

The study was emphasized on the applicability of Landsat data for vegetation classification of touch as small natural swamp areas Yujeon natural swamp in Haman-gun through image processing system. The image processing technique which was applied is maximum likelihood method. The classified types on the Landsat image are water, nelumbo, grass, agricultural field and conifer. The computer processed classification was compared the existed data for evaluating the result, but there are some difficulties on the exact comparison between them because of discordance of the temporal resolution. The result, anyhow, is quite remarkable that Landsat MSS data can be used for the quantitative estimates of vegetation type classification in such small area.

• Journal of information and communication convergence engineering
• /
• v.16 no.3
• /
• pp.173-178
• /
• 2018

Accurate classification of cloud images is a challenging task. Almost all the existing methods rely on hand-crafted feature extraction. Their limitation is low discriminative power. In the recent years, deep learning with convolution neural networks (CNNs), which can auto extract features, has achieved promising results in many computer vision and image understanding fields. However, deep learning approaches usually need large datasets. This paper proposes a deep learning approach for classification of cloud image patches on small datasets. First, we design a suitable deep learning model for small datasets using a CNN, and then we apply data augmentation and dropout regularization techniques to increase the generalization of the model. The experiments for the proposed approach were performed on SWIMCAT small dataset with k-fold cross-validation. The experimental results demonstrated perfect classification accuracy for most classes on every fold, and confirmed both the high accuracy and the robustness of the proposed model.

• Journal of Korean Society for Geospatial Information Science
• /
• v.17 no.2
• /
• pp.55-60
• /
• 2009

By applying NDVI(Normalized Difference Vegetation Index) and TCT(Tasseled-Cap Transformation) image algorithm on the basis of KOMSAP-2 MSC(Multi Spectral Camera) image(Jun. 12, 2007) for Naenam-myeon, Gyeongju city in this study, DN distribution map was drawn up. Discrimination analysis of image algorithm for the accuracy improvement of forest type classification was conducted through the comparative analysis between the distribution maps of NDVI and TCT DN, and forest field surveying data, and finally, the accuracy of the forest type classification was verified through the overlay analysis with the forest field surveying data. Through this study, it is thought that low cost and high efficiency will be able to be expected in the process of the examination for the automation practicality of the forest type classification and of the production of the accurate forest type classification map by using KOMPSAT-2 MSC image.

• Journal of Information Processing Systems
• /
• v.19 no.1
• /
• pp.67-79
• /
• 2023

In the image field, data augmentation refers to increasing the amount of data through an editing method such as rotating or cropping a photo. In this study, a generative adversarial network (GAN) image was created using CycleGAN, and various colors of dogs were reflected through data augmentation. In particular, dog data from the Stanford Dogs Dataset and Oxford-IIIT Pet Dataset were used, and 10 breeds of dog, corresponding to 300 images each, were selected. Subsequently, a GAN image was generated using CycleGAN, and four learning groups were established: 2,000 original photos (group I); 2,000 original photos + 1,000 GAN images (group II); 3,000 original photos (group III); and 3,000 original photos + 1,000 GAN images (group IV). The amount of data in each learning group was augmented using existing data augmentation methods such as rotating, cropping, erasing, and distorting. The augmented photo data were used to train the MobileNet_v3_Large, ResNet-152, InceptionResNet_v2, and NASNet_Large frameworks to evaluate the classification accuracy and loss. The top-3 accuracy for each deep neural network model was as follows: MobileNet_v3_Large of 86.4% (group I), 85.4% (group II), 90.4% (group III), and 89.2% (group IV); ResNet-152 of 82.4% (group I), 83.7% (group II), 84.7% (group III), and 84.9% (group IV); InceptionResNet_v2 of 90.7% (group I), 88.4% (group II), 93.3% (group III), and 93.1% (group IV); and NASNet_Large of 85% (group I), 88.1% (group II), 91.8% (group III), and 92% (group IV). The InceptionResNet_v2 model exhibited the highest image classification accuracy, and the NASNet_Large model exhibited the highest increase in the accuracy owing to data augmentation.

• Journal of the Korean Society of Surveying, Geodesy, Photogrammetry and Cartography
• /
• v.30 no.6_2
• /
• pp.653-663
• /
• 2012

Using soft classification, it is possible to obtain the land cover proportions from the remotely sensed image. These land cover proportions are then used as input data for a procedure called "super-resolution mapping" to produce the predicted hard land cover layers at higher resolution than the original remotely sensed image. Superresolution mapping can be implemented using a number of algorithms in which the Hopfield Neural Network (HNN) has showed some advantages. The HNN has improved the land cover classification through superresolution mapping greatly with the high resolution data. However, the super-resolution mapping is based on the spatial dependence assumption, therefore it is predicted that the accuracy of resulted land cover classes depends on the relative size of spatial features and the spatial resolution of the remotely sensed image. This research is to evaluate the capability of HNN to implement the super-resolution mapping for SPOT image to create higher resolution land cover classes with different zoom factor.

• Journal of the Korea Institute of Military Science and Technology
• /
• v.27 no.3
• /
• pp.319-328
• /
• 2024

Semi-supervised learning is a good way to train a classification model using a small number of labeled and large number of unlabeled data. We applied semi-supervised learning to a synthetic aperture radar(SAR) image classification model with a limited number of datasets that are difficult to create. To address the previous difficulties, semi-supervised learning uses a model trained with a small amount of labeled data to generate and learn pseudo labels. Besides, a lot of number of papers use a single fixed threshold to create pseudo labels. In this paper, we present a semi-supervised synthetic aperture radar(SAR) image classification method that applies different thresholds for each class instead of all classes sharing a fixed threshold to improve SAR classification performance with a small number of labeled datasets.

• ETRI Journal
• /
• v.38 no.1
• /
• pp.174-184
• /
• 2016

Information identification with image data by means of low-level visual features has evolved as a challenging research domain. Conventional text-based mapping of image data has been gradually replaced by content-based techniques of image identification. Feature extraction from image content plays a crucial role in facilitating content-based detection processes. In this paper, the authors have proposed four different techniques for multiview feature extraction from images. The efficiency of extracted feature vectors for content-based image classification and retrieval is evaluated by means of fusion-based and data standardization-based techniques. It is observed that the latter surpasses the former. The proposed methods outclass state-of-the-art techniques for content-based image identification and show an average increase in precision of 17.71% and 22.78% for classification and retrieval, respectively. Three public datasets - Wang; Oliva and Torralba (OT-Scene); and Corel - are used for verification purposes. The research findings are statistically validated by conducting a paired t-test.

• Journal of Internet Computing and Services
• /
• v.11 no.3
• /
• pp.15-24
• /
• 2010

In this paper, we propose an image classification which combines semantic relations of tags with contents of images to improve the satisfaction of image retrieval on application domains as huge image sharing sites. To make good use of image retrieval or classification algorithms on huge image sharing sites as Flickr, they are applicable to real tagged Web images. To classify the Web images by 'bag of visual word' based image content, our algorithm includes training the category model by utilizing the preliminary retrieved images with semantically related tags as training data and classifying the test images based on PLSA. In the experimental results on the Flickr Web images, the proposed method produced the better precision and recall rates than those from the existing method using tag information.

• KSCE Journal of Civil and Environmental Engineering Research
• /
• v.28 no.4D
• /
• pp.569-577
• /
• 2008

The purpose of this study is focused on the development of compound classification process by mixing multitemporal data and annexing a specific image enhancement technique with a specific image classification algorithm, to gain more accurate land information from satellite imagery. That is, this study suggests the classification process using canonical correlation classification technique after principal component analysis for the mixed multitemporal data. The result of this proposed classification process is compared with the canonical correlation classification result of one date images, multitemporal imagery and a mixed image after principal component analysis for one date images. The satellite images which are used are the Landsat 5 TM images acquired on July 26, 1994 and September 1, 1996. Ground truth data for accuracy assessment is obtained from topographic map and aerial photograph, and all of the study area is used for accuracy assessment. The proposed compound classification process showed superior efficiency to appling canonical correlation classification technique for only one date image in classification accuracy by 8.2%. Especially, it was valid in classifying mixed urban area correctly. Conclusively, to improve the classification accuracy when extracting land cover information using Landsat TM image, appling canonical correlation classification technique after principal component analysis for multitemporal imagery is very useful.