Shiuan Wan

Taiwan is located at the junction of plates in which the stratum is relatively unstable, resulting in frequent earthquakes. Driftwood has always been regarded as a precious asset that enables ecoscientists to track earthquakes. In the event of a typhoon or heavy rain, the surface water flows to flush the woods from the hills to the coast. More specifically, a large rainfall or earthquake may cause floods and collapses, and the trees in the forest will be washed down. Therefore, this study used high-resolution images to build an image database of the new north coast of Taiwan, and a deep learning approach is incorporated to classify the driftwoods. To improve the interpretation of driftwood in the remote images, we initially import eight pieces of textured information which are employed to the raw bands (B, G, R, and IR). The usage of spatial information image extraction technology is incorporated into a deep learning analysis using two parallel approaches. The generative adversarial network (GAN) is used to analyze the color images alongside an ancillary image with texture information. Most of the salt–pepper effects are produced by applying a high-resolution thematic map, and an error matrix is generated to compare the differences between them. The raw data (original R + G + B + IR) images, when analyzed using GAN, have about 70% overall classification outcomes. Not all of the driftwood can be detected. By applying the texture information to the parallel approach, the overall accuracy is enhanced to 78%, and about 80% of the driftwood can be recognized.

Currently, the seashore is threatened by the environment of climate change and increasing coastal waste. The past environmental groups used a large amount of manpower to manage the coast to maintain the seashore environment. The computational time cost and efficiency are not ideal for the vast area of the seashore. With the progress of GIS (Geographic Information System) technology, the ability of remote sensing technology can capture a wide range of data in a short period. This research is based on the application of remote sensing technology combined with machine learning to display the observation of our seashore. However, in the process of image classification, the seashore wastes are small, which required the use of high-resolution image data. Thus, how to remove the noise becomes a crucial issue in developing an image classifier machine. The difficulties include how to adjust the value of parameters for removing/avoiding noises. First, the texture information and vegetation indices were employed as ancillary information in our image classification. On the other hand, auto-encoder is a very good tool to denoise a given image; hence, it is used to transform high-resolution images by considering ancillary information to extract attributes. Multi-layer perceptron (MLP) and support vector machine (SVM) were compared for classifier performance in a parallel study. The overall accuracy is about 85.5% and 83.9% for MLP and SVM, respectively. If the AE is applied for preprocessing, the overall accuracy is increased by about 10–12%.

Water treatment is an important process, as it improves water quality and makes it better for any end use, whether it be drinking, industrial use, irrigation, water recreation, or any other kind of use. Turbidity is one of the fundamental measurements of the clarity of water in water treatment. Specifically, this component is an optical feature of the amount of light on scatter particles when light is shined on a water sample. It is crucial in water reservoirs to provide clean water, which is difficult to manage and predict. Hence, this study focuses on the use of robust deep learning models to analyze time-series data in order to predict the water quality of turbidity in a reservoir area. Deep learning models may become an alternative solution in predicting water quality because of their accuracy. This study is divided into two parts: (a) the first part uses the optical bands of blue (B), green (G), red (R), and infrared (IR) to build a regression function to monitor turbidity in water, and (b) the second part uses a hybrid model to analyze time-series turbidity data with the recursive neural network (RNN2) model. The selected models’ accuracies are compared based on the accuracy using the input data, forecasting level, and training time. The analysis shows that these models have their strengths and limitations under different analyzed conditions. Generally, RNN2 shows the performance regarding the root-mean-square error (RMSE) evaluation metric. The most significant finding is that the RNN2 model is suitable for the accurate prediction of water quality. The RMSE is used to facilitate a comparison of the accuracy of the sampling data. In the training model, the training data have an RMSE of 20.89, and the testing data have an RMSE of 30.11. The predicted R-squared values in the RNN2 model are 0.993 (training data) and 0.941 (testing data).

This study employed a data fusion method to extract the high-similarity time series feature index of a dataset through the integration of MS (Multi-Spectrum) and SAR (Synthetic Aperture Radar) images. The farmlands are divided into small pieces that consider the different behaviors of farmers for their planting contents in Taiwan. Hence, the conventional image classification process cannot produce good outcomes. The crop phenological information will be a core factor to multi-period image data. Accordingly, the study intends to resolve the previous problem by using three different SPOT6 satellite images and nine Sentinel-1A synthetic aperture radar images, which were used to calculate features such as texture and indicator information, in 2019. Considering that a Dynamic Time Warping (DTW) index (i) can integrate different image data sources, (ii) can integrate data of different lengths, and (iii) can generate information with time characteristics, this type of index can resolve certain classification problems with long-term crop classification and monitoring. More specifically, this study used the time series data analysis of DTW to produce “multi-scale time series feature similarity indicators”. We used three approaches (Support Vector Machine, Neural Network, and Decision Tree) to classify paddy patches into two groups: (a) the first group did not apply a DTW index, and (b) the second group extracted conflict predicted data from (a) to apply a DTW index. The outcomes from the second group performed better than the first group in regard to overall accuracy (OA) and kappa. Among those classifiers, the Neural Network approach had the largest improvement of OA and kappa from 89.51, 0.66 to 92.63, 0.74, respectively. The rest of the two classifiers also showed progress. The best performance of classification results was obtained from the Decision Tree of 94.71, 0.81. Observing the outcomes, the interference effects of the image were resolved successfully by various image problems using the spectral image and radar image for paddy rice classification. The overall accuracy and kappa showed improvement, and the maximum kappa was enhanced by about 8%. The classification performance was improved by considering the DTW index.

Remote sensing technology has rendered lots of information in agriculture. It has usually been used to monitor paddy growing ecosystems in the past few decades. However, there are uncertainties in data fusion techniques which can be resolved in image classification on paddy rice. In this study, a series of learning concepts integrated by a probability progress Fuzzy Dempster-Shafer (FDS) analysis is presented to upgrade various models and different types of image data which is the goal of this study. More specifically, the study utilized the FDS to generate a series of probability models in the classification of the system. In addition, Logistic Regression (LR), Support Vector Machine (SVM), and Neural Network (NN) approaches are employed into the developed FDS system. Furthermore, two different image types are Satellite Image and Aerial Photo used as the analysis material. The overall classification accuracy has been improved to 97.27%, and the kappa value is 0.93. The overall accuracy of the paddy field image classification for a multi-period of mid-scale satellite images is between 85% and 90%. The overall accuracy of the classification using multi-spectral numerical aerial photos can be between 91% and 95%. The FDS improves the accuracy of the above image classification results.

The analysis, measurement, and computation of remote sensing images often require enhanced unsupervised/supervised classification approaches. The goal of this study is to have a better understanding of (a) the classification performance of multispectral image and hyperspectral image data; (b) the classification performance of unsupervised and supervised models; and (c) the classification performance of feature selection among different models. More specifically, the multispectral images and hyperspectral images with high spatial resolution are well accepted for improving land use and classification. Hence, this study used multispectral images (WorldView-2) and hyperspectral images (CASI-1500) and focused on the classifiers K-means, density-based spatial clustering of applications with noise (DBSCAN), linear discriminant analysis (LDA), and back-propagation neural network (BPN). Then the feature selection (principle component analysis, PCA) on four classifiers is studied. The results show that the image material of CASI-1500 classification accuracy is slightly better than that of WorldView-2. The overall classification of BPN is the best, the overall data has a κ value of 0.89 and the overall accuracy is 97%. The DBSCAN presents a reality with good accuracy and the integrity of the thematic map. The DBSCAN can attain an accuracy of about 88% and save 95.1% of computational time.

Landslides of Taiwan occur frequently in high mountain areas. Soil disturbance causes by the earthquake and heavy rainfall of the typhoon seasons often produced the earth and rock to landslide in the upper reaches of the catchment area. Therefore, the landslide near the hillside has an influence on the catchment area. The hyperspectral images are effectively used to monitor the landslide area with the spectral analysis. However, it is rarely studied how to interpret it in the image of the landslide. If there are no elevation data on the slope disaster, it is quite difficult to identify the landslide zone and the bareland area. More specifically, this study used a series of spectrum analysis to identify the difference between them. Therefore, this study conducted a spectrum analysis for the classification of the landslide, bareland, and vegetation area in the mountain area of NanXi District, Tainan City. On the other hand, this study used the following parallel study on Support Vector Machine (SVM) for error matrix and thematic map for comparison. The study simultaneously compared the differences between them. The spectral similarity analysis reaches 85% for testing data, and the SVM approach has 98.3%.