Semantic segmentation of land cover from high resolution multispectral satellite images by spectral-spatial convolutional neural network

In classification of satellite images acquired over smallholder agricultural landscape with complex spectral profiles of various crop types, exploring image spatial information is important. The deep convolutional neural network (CNN), originally designed for natural image recognition in the computer vision field, can automatically explore high level spatial information and thus is promising for such tasks. This study tried to evaluate different CNN structures for classification of four smallholder agricultural landscapes in Heilongjiang, China using pan-sharpened 2 m GaoFen-1 (meaning high resolution in Chinese) satellite images. CNN with three pooling strategies: without pooling, with max pooling and with average pooling, were evaluated and compared with random forest. Two different numbers (~70,000 and ~290,000) of CNN learnable parameters were examined for each pooling strategy. The training and testing samples were systematically sampled from reference land cover maps to ensure sample distribution proportional to the reference land cover occurrence and included 60,000–400,000 pixels to ensure effective training. Testing sample classification results in the four study areas showed that the best pooling strategy was the average pooling CNN and that the CNN significantly outperformed random forest (2.4–3.3% higher overall accuracy and 0.05–0.24 higher kappa coefficient). Visual examination of CNN classification maps showed that CNN can discriminate better the spectrally similar crop types by effectively exploring spatial information. CNN was still significantly outperformed random forest using training samples that were evenly distributed among classes. Furthermore, future research to improve CNN performance was discussed.

Download Full-text

OIL PATTERN IDENTIFICATION ANALYSIS USING SEMANTIC DEEP LEARNING METHOD FROM PLEIADES-1B SATELIITE IMAGERY WITH ARCGIS PRO SOFTWARE (Case Study: Village “A”)

IOP Conference Series Earth and Environmental Science ◽

10.1088/1755-1315/936/1/012021 ◽

2021 ◽

Vol 936 (1) ◽

pp. 012021

Author(s):

Novi Anita ◽

Bangun Muljo Sukojo ◽

Sondy Hardian Meisajiwa ◽

Muhammad Alfian Romadhon

Keyword(s):

Neural Network ◽

Deep Learning ◽

High Resolution ◽

Land Cover ◽

Convolutional Neural Network ◽

Local Community ◽

Confusion Matrix ◽

Test Method ◽

Oil Refinery ◽

Thermal Data

Abstract There are many petroleum mining activities scattered in developing countries, such as Indonesia. Indonesia is one of the largest oil-producing countries in Southeast Asia with the 23rd ranking. Since the Dutch era, Indonesia has produced a very large amount of petroleum. One of the oil producing areas is “A” Village. There is an old well that produces petroleum oil which is still active with an age of more than 100 years, for now the oil well is still used by the local community as the main source of livelihood. With this activity, resulting in an oil pattern around the old oil refinery, which over time will absorb into the ground. This study aims to analyze and identify the oil pattern around the old oil refinery in the “A” area. The data used is in the form of High-Resolution Satellite Imagery (CSRT), namely Pleiades-1B with a spatial resolution of 1.5 meters. Data were identified using the Deep Learning Semantic method. For the limitation of this research is the administrative limit of XX Regency with a scale of 1: 25,000 as supporting data when cutting the image. The method used is the Deep Learning Convolutional Neural Network series. This research is based on how to wait for the method of the former oil spill which is the consideration of the consideration used. This study produced a land cover map that was classified into 3 categories, namely oil patterns area, area not affected by oil and vegetation. As a supporting value to show the accuracy of the classification results, an accuracy test method is used with the confusion matrix method. To show the accuracy of this study using thermal data taken from the field. Thermal data used in the form of numbers that show the temperature of each land cover. Based on the above reference, a research related to the analysis of very high-resolution image data (Pleiades-1B) will be conducted to examine the oil pattern. This research uses the deep learning series convolutional neural network (CNN) method. With this research, it is hoped that it can help agencies in knowing the right method to identify oil in mainland areas.

Download Full-text

Tree Crown Delineation Algorithm Based on a Convolutional Neural Network

Remote Sensing ◽

10.3390/rs12081288 ◽

2020 ◽

Vol 12 (8) ◽

pp. 1288 ◽

Cited By ~ 5

Author(s):

José R. G. Braga ◽

Vinícius Peripato ◽

Ricardo Dalagnol ◽

Matheus P. Ferreira ◽

Yuliya Tarabalka ◽

...

Keyword(s):

Neural Network ◽

High Resolution ◽

Convolutional Neural Network ◽

Tropical Forests ◽

Large Scale ◽

Satellite Images ◽

Tree Level ◽

Tree Crown ◽

Individual Tree ◽

High Resolution Satellite Images

Tropical forests concentrate the largest diversity of species on the planet and play a key role in maintaining environmental processes. Due to the importance of those forests, there is growing interest in mapping their components and getting information at an individual tree level to conduct reliable satellite-based forest inventory for biomass and species distribution qualification. Individual tree crown information could be manually gathered from high resolution satellite images; however, to achieve this task at large-scale, an algorithm to identify and delineate each tree crown individually, with high accuracy, is a prerequisite. In this study, we propose the application of a convolutional neural network—Mask R-CNN algorithm—to perform the tree crown detection and delineation. The algorithm uses very high-resolution satellite images from tropical forests. The results obtained are promising—the R e c a l l , P r e c i s i o n , and F 1 score values obtained were were 0.81 , 0.91 , and 0.86 , respectively. In the study site, the total of tree crowns delineated was 59,062 . These results suggest that this algorithm can be used to assist the planning and conduction of forest inventories. As the algorithm is based on a Deep Learning approach, it can be systematically trained and used for other regions.

Download Full-text