Object-oriented Change Detection Method Research Articles

A Novel UNet 3+ Change Detection Method Considering Scale Uncertainty in High-Resolution Imagery

The challenge of detecting changes in high-resolution remote sensing imagery often stems from the difficulties in effectively extracting features and constructing appropriate change detection models considering the scale characteristics of ground objects. To solve these issues, we propose a novel UNet 3+ change detection method that considers the scale characteristics inherent in various land-cover change types. Our method includes three key steps: a multi-scale segmentation method, a class-specific UNet 3+ method, and an object-oriented change detection method based on UNet 3+. To verify the effectiveness of this method, we select two datasets for experiments and compare our proposed method with the UNet 3+ single-scale sampling method, the class-specific UNet 3+ single-scale sampling method, and the UNet 3+ multi-scale hierarchical sampling method. Our experimental results show that our proposed method has higher overall accuracy and F1, lower missed detection rate and false detection rate, and can detect more changes in ground features than other methods. To verify the scalability of this method, we compare this method with traditional change detection methods such as PCA-k-means, OCVA, a single-scale sampling method based on random forest, and a class-specific object-based method. Experimental results and accuracy indexes show that our proposed method better considers the scale characteristics of ground objects and achieves higher accuracy. Additionally, we compared our proposed method with other DLCD methods including LamboiseNet, BIT, CDNet, FCSiamConc, and FCSiamDiff. Our results show that our proposed method effectively considers edge information and has an acceptable time consumption. Our approach not only considers the full-scale characteristics of the feature extraction but also the scale characteristics of the change detection model. In addition, it considers a more practical feature extraction unit (object), making it more accurate.

Object-Oriented Unsupervised Change Detection Based on Neighborhood Correlation Images and k-Means Clustering for the Multispectral and High Spatial Resolution Images

An unsupervised change-detection problem is formulated as a binary classification problem corresponding to the change and no change areas. This paper proposes a novel unsupervised object-oriented change detection method based on neighborhood correlation images (NCIs) and k-means clustering for high-resolution remote sensing images. We tested our proposed method in two study areas of Beijing with RapidEye images and compared it with three other popular change detection methods based on different images: change vector analysis (CVA), principal component analysis (PCA), and multivariate alteration detection (MAD). The results indicate that our method has the highest overall accuracy (90.80% in Shunyi District, Beijing and 90.40% in Daxing District, Beijing) and Kappa coefficient (0.7922 in Shunyi District, Beijing and 0.7796 in Daxing District, Beijing). In addition, the McNemar test indicates that our method is robust and stable across different study areas. We concluded that the object-oriented NCIs method outperforms traditional difference images (CVA, PCA, and MAD) in unsupervised change detection. The experimental results demonstrate the effectiveness of the proposed approach in solving the problem of unsupervised change detection for high-resolution images.

A Multi-Feature Fusion-Based Change Detection Method for Remote Sensing Images

An object-oriented change detection method for remote sensing images based on multiple features using a novel weighted fuzzy c-means (WFCM) method is presented. First, Gabor and Markov random field textures are extracted and added to the original images. Second, objects are obtained by using a watershed segmentation algorithm to segment the images. Third, simple threshold technology is applied to produce the initial change detection results. Finally, refining is conducted using WFCM with different feature weights identified by the Relief algorithm. Two satellite images are used to validate the proposed method. Experimental results show that the proposed method can reduce uncertainties involved in using a single feature or using equally weighted features, resulting in higher accuracy.

Object-oriented change detection method based on adaptive multi-method combination for remote-sensing images

ABSTRACTIn this study, we propose a novel object-oriented change detection method for remote-sensing images. First, the Gabor texture and Markov random field texture are extracted based on the remote-sensing images, and an initial pixel-level change detection result is produced. Second, in order to reduce the influence of feature uncertainty on the change detection results, the weights of different features are calculated by the Relief algorithm based on the initial pixel-level change detection result, and several difference images are fused to obtain a single comprehensive difference image. Third, different pixel-level change detection results are obtained using diverse change detection methods. The two-temporal images are then stacked and segmented, and to ensure change detection method separability, the weighted object change probability is obtained by fusing five different object change probabilities, which are calculated from the pixel-level change detection results. Finally, the objects are labelled as the class with a higher weighted object change probability. Our experimental results showed that the accuracy of change detection results obtained using the weighted object change probability was higher than that of the change detection results produced using the independent object change probability.

OBJECT-ORIENTED CHANGE DETECTION BASED ON MULTI-SCALE APPROACH

The change detection of remote sensing images means analysing the change information quantitatively and recognizing the change types of the surface coverage data in different time phases. With the appearance of high resolution remote sensing image, object-oriented change detection method arises at this historic moment. In this paper, we research multi-scale approach for high resolution images, which includes multi-scale segmentation, multi-scale feature selection and multi-scale classification. Experimental results show that this method has a stronger advantage than the traditional single-scale method of high resolution remote sensing image change detection.

OBJECT-ORIENTED CHANGE DETECTION BASED ON MULTI-SCALE APPROACH

Abstract. The change detection of remote sensing images means analysing the change information quantitatively and recognizing the change types of the surface coverage data in different time phases. With the appearance of high resolution remote sensing image, object-oriented change detection method arises at this historic moment. In this paper, we research multi-scale approach for high resolution images, which includes multi-scale segmentation, multi-scale feature selection and multi-scale classification. Experimental results show that this method has a stronger advantage than the traditional single-scale method of high resolution remote sensing image change detection.

COMPARISON OF PIXEL-BASED AND OBJECT-ORIENTED LAND COVER CHANGE DETECTION METHODS

Abstract. Change detection method is an efficient way in the aim of land cover product updating on the basis of the existing products, and at the same time saving lots of cost and time. Considering the object-oriented change detection method for 30m resolution Landsat image, analysis of effect of different segmentation scales on the method of the object-oriented is firstly carried out. On the other hand, for analysing the effectiveness and availability of pixel-based change method, the two indices which complement each other are the differenced Normalized Difference Vegetation Index (dNDVI), the Change Vector (CV) were used. To demonstrate the performance of pixel-based and object-oriented, accuracy assessment of these two change detection results will be conducted by four indicators which include overall accuracy, omission error, commission error and Kappa coefficient.

COMPARISON OF PIXEL-BASED AND OBJECT-ORIENTED LAND COVER CHANGE DETECTION METHODS

Change detection method is an efficient way in the aim of land cover product updating on the basis of the existing products, and at the same time saving lots of cost and time. Considering the object-oriented change detection method for 30m resolution Landsat image, analysis of effect of different segmentation scales on the method of the object-oriented is firstly carried out. On the other hand, for analysing the effectiveness and availability of pixel-based change method, the two indices which complement each other are the differenced Normalized Difference Vegetation Index (dNDVI), the Change Vector (CV) were used. To demonstrate the performance of pixel-based and object-oriented, accuracy assessment of these two change detection results will be conducted by four indicators which include overall accuracy, omission error, commission error and Kappa coefficient.

Object-oriented change detection based on weighted polarimetric scattering differences on POLSAR images

Abstract. For wide application of change detection with SAR imagery, current processing technologies and methods are mostly based on pixels. It is difficult for pixel-based technologies to utilize spatial characteristics of images and topological relations of objects. Object-oriented technology takes objects as processing unit, which takes advantage of the shape and texture information of image. It can greatly improve the efficiency and reliability of change detection. Recently, with the development of polarimetric synthetic aperture radar (PolSAR), more backscattering features on different polarization state can be available for usage of object-oriented change detection study. In this paper, the object-oriented strategy will be employed. Considering the fact that the different target or target's state behaves different backscattering characteristics dependent on polarization state, an object-oriented change detection method that based on weighted polarimetric scattering difference of PolSAR images is proposed. The method operates on the objects generated by generalized statistical region merging (GSRM) segmentation processing. The merit of GSRM method is that image segmentation is executed on polarimetric coherence matrix, which takes full advantages of polarimetric backscattering features. And then, the measurement of polarimetric scattering difference is constructed by combining the correlation of covariance matrix and the difference of scattering power. Through analysing the effects of the covariance matrix correlation and the scattering echo power difference on the polarimetric scattering difference, the weighted method is used to balance the influences caused by the two parts, so that more reasonable weights can be chosen to decrease the false alarm rate. The effectiveness of the algorithm that proposed in this letter is tested by detection of the growth of crops with two different temporal radarsat-2 fully PolSAR data. First, objects are produced by GSRM algorithm based on the coherent matrix in the pre-processing. Then, the corresponding patches are extracted in two temporal images to measure the differences of objects. To detect changes of patches, a difference map is created by means of weighted polarization scattering difference. Finally, the result of change detection can be obtained by threshold determining. The experiments show that this approach is feasible and effective, and a reasonable choice of weights can improve the detection accuracy significantly.

基于多尺度分割和决策树算法的山区遥感影像变化检测方法——以四川攀西地区为例

In mountainous area,many factors such as the high spatial heterogeneity in land cover conditions and the topographic effect on remote sensing data acquirement significantly constraint the change detection by using remote sensing method. It has become a great challenge to develop an effective method to accurately map the land cover changes for complexterrain areas. In the traditional change detection methods,the spectral features are the major information. Due to the lack of the consideration of other information from structure, topography and other parameters, many false changes will be introduced in the final detection results with a big uncertainty. To avoid the false change and overcome the influence from terrain shadow and phenology differences, an object-oriented change detection method by combining multi-scale segmentation and decision tree algorithm for complex terrain condition is proposed in this study. In the new method,firstly,a series of spectrum information( i.e. reflectance,change vector intensity,NDVI),shape information( i. e. area,lengthwidth ratio,shape index) and terrain parameters( i.e. elevation,slope,aspect) are calculated for change detection rules construction. Secondly,the unchanged surfaces which are usually detected as the false changes due to the influence of terrain shadow and phenology difference are extracted as training sample. Finally,the change detection rules are built automatically by C5.0 decision tree algorithm. the proposed method is applied to Landsat-TM images acquired in 1989 and2009 at Panxi region,the detection results indicate:( 1) Over the mountainous area,the method can effectively reduce the influence from topographic effect,and has a big improvement on detection accuracy with the overall accuracy of 93.57% and Kappa coefcient of 0.8706 by validating ground reference samples and stratified random samples.( 2) The C5.0 decision tree algorithm performs well in terms of robustness and adaptability,which serves well in land cover classification with types only labeled as "changed "and "unchanged "and highly heterogeneous training samples. Considering the false changes result from topographic and phenological as "unchanged "training sample,the accuracy of change detection is improved effectively.( 3) Spectral features are the predominant information in analyzing TM images,and the NDVI differences between the reference image and detected image is an effective indicator for change detection of land cover pattern in vegetation coverage region.( 4) The changes of land cover pattern in Panxi region from 1989 to 2009 were closely associated with human activities. The typical driving forces of pattern change include returning farmland to forest policy,water conservancy project,mining,urban sprawl and so on. The changed area amounted to 740.2 km2,accounting for 2.49% of the total detected area.

The change detection method based on object-oriented characteristic mapping pattern analysis

This paper proposes a object-oriented change detection method based on the characteristic mapping pattern analysis.The method is improved from the information transfer model of remote sensing image interpretation.The image objects are acquired by the vector auxiliary data.The spectral and texture features are extracted,and an unsupervised clustering method is used to obtain the characteristic clusters of the objects.According to the priori information which exists in the auxiliary data,the mapping between the multi-temporal feature clusters is analyzed class by class.Then,the change object,whose mapping mode is inconsistent with other objects of the same class,can be identified.The experiments prove the feasibility and effectiveness of the proposed method,and the results show a new way for the object-oriented change detection.

Road Change Detection Using Object-oriented Projective Interactive Partition

In this paper the authors present an object-oriented change detection method which uses projective interactive partition to detect the change of the urban road.The QuickBird images acquired in different years were used in the experiment,on which multi-scale segmentation and hierarchical classification were carried out for urban road extraction according to the spectrum,shape and texture features.Then the object layers were established in the same detection,and they were the projection layers taking over the classification results of asynchrony images.Interactive partition was also realized on these layers.Finally the change detection results were achieved after judging the consistency of land feature categories.The experiment results show that the method of object-oriented change detection using projective interactive partition can extract and detect the urban road effectively.