Comparison of GEOBIA classification algorithms based on Worldview-3 imagery in the extraction of coastal coniferous forest

Šume primorskih četinjača, sa svojom ekološkom, ekonomskom, estetskom i društvenom funkcijom, predstavljaju važan dio europskih šumskih zajednica. Osnovni cilj ovoga rada je usporediti najkorištenije GEOBIA (engl. Geographic Object-Based Image Analysis) klasifikacijske algoritme (engl. Random Trees – RT, Maximum Likelihood – ML, Support Vector Machine – SVM) s ciljem izdvajanja šuma primorskih četinjača na visoko-rezolucijskom WorldView-3 snimku unutar topografskog slijevnog područja naselja Split. Metodološki okvir istraživanja uključuje (1) izvođenje izoštrenog multispektralnog snimka (WV-3_MS-a); (2) testiranje segmentacijskih korisničko-definiranih parametara; (3) dodavanje testnih uzoraka; (4) klasifikaciju segmentiranog modela; (5) procjenu točnosti klasifikacijskih algoritama, te (6) procjenu točnosti završnog modela. RT se prema korištenim pokazateljima (correctness – COR, completeness – COM i overall quality – OQ) pokazao kao najbolji algoritam. Iterativno postavljanje segmentacijskih parametara omogućilo je detekciju najprikladnijih vrijednosti za generiranje segmentacijskog modela. Utvrđeno je da sjene mogu uzrokovati značajne probleme ako se klasificiranje vrši na visoko-rezolucijskim snimkama. Modificiranim Cohen’s kappa coefficient (K) pokazateljem izračunata je točnost konačnog modela od 87,38%. WV-3_MS se može smatrati kvalitetnim podatkom za detekciju šuma primorskih četinjača primjenom GEOBIA metode.

Fast Urban Land Cover Mapping Exploiting Sentinel-1 and Sentinel-2 Data

The rapid change and expansion of human settlements raise the need for precise remote-sensing monitoring tools. While some Land Cover (LC) maps are publicly available, the knowledge of the up-to-date urban extent for a specific instance in time is often missing. The lack of a relevant urban mask, especially in developing countries, increases the burden on Earth Observation (EO) data users or requires them to rely on time-consuming manual classification. This paper explores fast and effective exploitation of Sentinel-1 (S1) and Sentinel-2 (S2) data for the generation of urban LC, which can be frequently updated. The method is based on an Object-Based Image Analysis (OBIA), where one Multi-Spectral (MS) image is used to define clusters of similar pixels through super-pixel segmentation. A short stack (<2 months) of Synthetic Aperture Radar (SAR) data is then employed to classify the clusters, exploiting the unique characteristics of the radio backscatter from human-made targets. The repeated illumination and acquisition geometry allows defining robust features based on amplitude, coherence, and polarimetry. Data from ascending and descending orbits are combined to overcome distortions and decrease sensitivity to the orientation of structures. Finally, an unsupervised Machine Learning (ML) model is used to separate the signature of urban targets in a mixed environment. The method was validated in two sites in Portugal, with diverse types of LC and complex topography. Comparative analysis was performed with two state-of-the-art high-resolution solutions, which require long sensing periods, indicating significant agreement between the methods (averaged accuracy of around 90%).

DELIMITAÇÃO DE VOÇOROCAS POR MEIO DE GEOBIA (Geographic Object Based Image Analysis) E MINERAÇÃO DE DADOS

Nos tempos atuais, com as medidas ambientais em evidência, é comum a discussão sobre melhores maneiras de obter informações cartográficas sobre as áreas degradadas. já que essas, são de suma importância para a efetividade de alguns programas como: Cadastro Ambiental Rural (CAR) e o Programa de Recuperação Ambiental (PRA).As degradações do solo, do tipo voçorocas, apresentam diversos prejuízos à natureza, pois possui estado irreversível, sendo possível somente sua recuperação parcial.O monitoramento destas áreas, bem como informações sobre as mesmas, é de suma importância para que se possa garantir o controle e definir métodos de conservação. Sob esta problemática este trabalho visa avaliar a eficiência de um procedimento de classificação, orientada, semiautomática (GEOBIA) em produtos cartográficos produzidos por Aeronaves Remotamente Pilotadas (ARP) para a delimitação de voçorocas. A utilização dos produtos cartográficos oriundos de ARP (Modelo Digital de Elevação e Ortoimagem Digital) se justifica devido ao baixo custo da ferramenta, bem como a potencialidade planialtimétrica.Os procedimentos foram realizados em duas áreas de estudo,situadas no município de Itajubá-MG. Nestas áreas foram definidos alguns pontos de controle e checagem para a classificação dos produtos cartográficos em relação ao Padrão de Exatidão Cartográfico Digital (PEC-PCD). Os produtos apresentaram classe A, para a escala 1/2.000. Foram determinados alguns parâmetros de segmentação para que se formassem segmentos fidedignos para cada área de estudo em específico, em seguida determinaram-se os atributos mais relevantes para a classificação e confecção da árvore de decisão de cada área. Na confecção da árvore de decisão utilizou-se o algoritmo C4.5.Os resultados foram satisfatórios a níveis de precisão (índice Kappa entre 0,88 e 0,92), tornando possível que as técnicas utilizadas em produtos cartográficos oriundos de ARP sejam uma ferramenta para delimitação de áreas degradadas do tipo voçoroca

Recognition of Ziziphus lotus through Aerial Imaging and Deep Transfer Learning Approach

There is a growing demand for the detection of endangered plant species through machine learning approaches. Ziziphus lotus is an endangered deciduous plant species in the buckthorn family (Rhamnaceae) native to Southern Europe. Traditional methods such as object-based image analysis have achieved good recognition rates. However, they are slow and require high human intervention. Transfer learning-based methods have several applications for data analysis in a variety of Internet of Things systems. In this work, we have analyzed the potential of convolutional neural networks to recognize and detect the Ziziphus lotus plant in remote sensing images. We fine-tuned Inception version 3, Xception, and Inception ResNet version 2 architectures for binary classification into plant species class and bare soil and vegetation class. The achieved results are promising and effectively demonstrate the better performance of deep learning algorithms over their counterparts.

Benthic habitat mapping using Object-Based Image Analysis (OBIA) on Tidung Island, Kepulauan Seribu, DKI Jakarta

Tidung Island is one of the islands in Kepulauan Seribu, DKI Jakarta, Indonesia. This island has various benthic that live on the coastal areas, and benthic habitat has various functions both ecologically and economically. Nowadays, remote sensing technology is one way to detect benthic habitats in coastal areas. Mapping benthic habitat is essential for sustainable coastal resource management and to predict the distribution of benthic organisms. This study aims to map the benthic habitats using the object-based image analysis (OBIA) and calculate the accuracy of benthic habitat classification results in Tidung Island, Kepulauan Seribu, DKI Jakarta. The field data were collected on June 2021, and the image data used is satellite Sentinel-2 imagery acquired in June 2021. The result shows that the benthic habitat classification was produced in 4 classes: seagrass, rubble, sand, and live coral. The accuracy test result obtained an overall accuracy (OA) of 74.29% at the optimum value of the MRS segmentation scale 15;0,1;0.7 with the SVM algorithm. The results of benthic habitat classification show that the Seagrass class dominates the shallow water area at the research site with an area of 118.77 ha followed by Life Coral 104.809 ha, Sand 43.352 ha, and the smallest area is the Rubble class of 42.28 Ha.

Object-based image analysis (OBIA) for mapping mangrove using Unmanned Aerial Vehicle (UAV) on Tidung Kecil Island, Kepulauan Seribu, DKI Jakarta Province

Tidung Kecil Island is a conservation and mangrove cultivation area. Therefore, the potential of mangrove ecosystems on Tidung Kecil Island will have a direct role in coastal ecosystems. Accurate mangrove mapping is necessary for the effective planning and management of ecosystems and resources because mangroves function as protectors of ecological systems. The utilization of remote sensing technology that is near real-time can be used as an alternative in providing spatial data effectively. Mapping earth’s surface objects method is growing especially after the development of design, research, and production of flexible Unmanned Aerial Vehicle (UAV) platforms. The use of object-based classification methods is currently an alternative in classifying an object of the Earth’s surface using both satellite and aerial photo imagery data (orthophoto) that has a high accuracy value. This research aim is to map object based mangrove ecosystems using UAV technology on Tidung Kecil Island, Kepulauan Seribu, DKI Jakarta. The K-NN algorithm result was a good classification with 81.081% overall accuracy (OA) at the optimum value of the MRS segmentation scale 300;0,1;0.7 and divided into two classes which are mangrove and non-mangrove for 0.381 ha and 20.912 ha respectively.

SEMI-AUTOMATED DELINEATION OF INFORMAL SETTLEMENT STRUCTURES FROM DRONE RGB IMAGERY USING OBJECT-BASED IMAGE ANALYSIS

With the problem of informal settlements in the Philippines, mapping such areas is the first step towards improvement. Object-based image analysis (OBIA) has been a powerful tool for mapping and feature extraction, especially for high-resolution datasets. In this study, an informal settlement area in UP Diliman, Quezon City was chosen to be the subject site, where individual informal settlement structures (ISS) were delineated and estimated using OBIA. With the help of photogrammetry and image enhancement techniques, derivatives such as elevation model and orthophotos were produced for easier interpretation. An initial rule-set was developed to remove all non-ISS features from the base image–utilizing spectral values and thematic layers as main classifiers. This classification technique yielded a 94% accuracy for non-ISS class, and 92% for the possible ISS class. Another rule-set was then developed to delineate individual ISS based on the texture and elevation model of the area, which paved the way for the estimation of ISS count. To test the robustness of the methodology developed, the estimation results were compared to the manual count obtained through an online survey form, and the classification and delineation results were assessed through overall and individual quality checks. The estimation yielded a relative accuracy of 60%, which came from the delineation rate of 63%. On the other hand, delineation accuracy was calculated through area-based and number-based measures, yielding 58% and 95%, respectively. Issues such as noisy elevation models and physical limitations of the area and survey done affected the accuracy of the results.