scholarly journals A Multi-feature Optimization Approach to Object-Based Image Classification

2006 ◽  
pp. 310-319 ◽  
Author(s):  
Qianni Zhang ◽  
Ebroul Izquierdo
Keyword(s):  
Image Classification ◽  
Optimization Approach ◽  
Object Based ◽  
Feature Optimization

scholarly journals Identifikasi Gosong Karang Mengggunakan Citra Satelit Sentinel 2A (Studi Kasus: Perairan Pesisir Nias Utara)

2020 ◽  
Vol 21 (1) ◽  
pp. 95-108
Author(s):  
Anang Dwi Purwanto ◽  
Teguh Prayogo ◽  
Sartono Marpaung
Keyword(s):  
Coral Reefs ◽  
Image Classification ◽  
Classification Method ◽  
Fishing Ground ◽  
Field Observations ◽  
Geometric Correction ◽  
Radiometric Correction ◽  
Object Based ◽  
Sentinel 2A ◽  
North Sumatra

ABSTRACTThe waters of Northern Nias, North Sumatra Province have a great potential for natural resources, one of which is the reef which is often used as a fishing ground. This study aims to identify and monitor the distribution of coral reefs around the waters of Northern Nias. The location of study is limited by coordinates 97° 0'31'' - 97° 16'54'' E and 1° 29'2'' LU - 1° 6'24'' N. The study locations were grouped in 6 (six) areas including Mardika reef, Wunga reef, Mausi1 reef, Mausi2 reef, Tureloto reef and Senau reef. The data used were Sentinel 2A imagery acquisition on 19 September 2018 and field observations made on 6-12 September 2018. Data processing includes geometric correction, radiometric correction, water column correction and classification using pixel-based and object-based methods as well as by delineating on the image. One classification method will be chosen that is most suitable for the location of the reef. The results show Sentinel 2A was very helpful in mapping the distribution of coral reefs compared to direct observation in the field. The use of image classification method rightly is very helpful in distinguishing coral reef objects from surrounding objects. The estimated area of coral reefs was 1,793.20 ha with details of the Mardika reef 143.27 ha, Wunga reef 627.06 ha, Mausi1 reef 299.84 ha, Mausi2 reef 141.873 ha, Tureloto reef 244.73 ha, Senau reef 336.44 ha. The existence of coral reefs have a high potential as a fishing ground and a natural tourist attraction.Keywords: coral reefs, sentinel 2A, lyzenga 1978, image classification, Northern NiasABSTRAKPerairan Nias Utara yang terletak di Provinsi Sumatra Utara memiliki potensi kekayaan alam yang besar dimana salah satunya adalah gosong karang yang sering dijadikan lokasi penangkapan ikan oleh nelayan. Penelitian ini bertujuan untuk mengidentifikasi dan monitoring sebaran gosong karang di sekitar perairan Nias Utara. Lokasi penelitian dibatasi dengan koordinat 97°0’31’’ - 97°16’54’’ BT dan 1°29’2’’LU – 1°6’24’’  LU. Untuk mempermudah dalam pengolahan data maka lokasi kajian dikelompokkan dalam 6 (enam) kawasan diantaranya gosong Mardika, gosong Wunga, gosong Mausi1, gosong Mausi2, gosong Tureloto dan gosong Senau. Data yang digunakan adalah citra satelit Sentinel 2A hasil perekaman tanggal 19 September 2018 dan hasil pengamatan lapangan yang telah dilakukan pada tanggal 6 - 12 September 2018. Pengolahan data meliputi koreksi geometrik, koreksi radiometrik, koreksi kolom air dan klasifikasi menggunakan metode klasifikasi berbasis piksel dan berbasis objek serta deliniasi citra. Dari ketiga metode klasifikasi tersebut akan dipilih satu metode klasifikasi yang sesuai dengan lokasi gosong karang. Hasil penelitian menunjukkan citra Sentinel 2A sangat membantu dalam memetakan sebaran gosong karang dibandingkan dengan pengamatan langsung di lapangan. Pemilihan metode klasifikasi citra satelit yang tepat sangat membantu dalam membedakan objek gosong karang dengan objek di sekitarnya. Estimasi total luasan gosong karang di perairan Nias Utara adalah 1,793.20 ha dengan rincian luasan gosong karang Mardika 143.27 ha, gosong Wunga 627.06 ha, gosong Mausi1 299.84 ha, gosong Mausi2 141.873 ha, gosong Tureloto 244.73 ha, gosong Senau 336.44 ha. Keberadaan gosong karang memiliki potensi yang tinggi sebagai lokasi penangkapan ikan dan memiliki daya tarik sebagai tempat wisata alam.Kata kunci: gosong karang, sentinel 2A, lyzenga 1978, klasifikasi citra, Nias Utara

scholarly journals Hyperspectral Imagery Classification Based on Multiscale Superpixel-Level Constraint Representation

2020 ◽  
Vol 12 (20) ◽  
pp. 3342
Author(s):  
Haoyang Yu ◽  
Xiao Zhang ◽  
Meiping Song ◽  
Jiaochan Hu ◽  
Qiandong Guo ◽  
...  
Keyword(s):  
Image Classification ◽  
Satellite Data ◽  
Hyperspectral Image ◽  
Training Sample ◽  
Decision Fusion ◽  
Spatial Constraint ◽  
Hyperspectral Image Classification ◽  
Object Based ◽  
Sparse Coefficient ◽  
The Individual

Sparse representation (SR)-based models have been widely applied for hyperspectral image classification. In our previously established constraint representation (CR) model, we exploited the underlying significance of the sparse coefficient and proposed the participation degree (PD) to represent the contribution of the training sample in representing the testing pixel. However, the spatial variants of the original residual error-driven frameworks often suffer the obstacles to optimization due to the strong constraints. In this paper, based on the object-based image classification (OBIC) framework, we firstly propose a spectral–spatial classification method, called superpixel-level constraint representation (SPCR). Firstly, it uses the PD in respect to the sparse coefficient from CR model. Then, transforming the individual PD to a united activity degree (UAD)-driven mechanism via a spatial constraint generated by the superpixel segmentation algorithm. The final classification is determined based on the UAD-driven mechanism. Considering that the SPCR is susceptible to the segmentation scale, an improved multiscale superpixel-level constraint representation (MSPCR) is further proposed through the decision fusion process of SPCR at different scales. The SPCR method is firstly performed at each scale, and the final category of the testing pixel is determined by the maximum number of the predicated labels among the classification results at each scale. Experimental results on four real hyperspectral datasets including a GF-5 satellite data verified the efficiency and practicability of the two proposed methods.

Sign in / Sign up

Export Citation Format

Share Document