RESEARCH ARTICLE Selection of Optimal Object Features in Object-Based Image Analysis Using Filter-Based Algorithms Ismail Colkesen 1 • Taskin Kavzoglu 1 Ó Indian Society of Remote Sensing 2018 Abstract With the increase in spatial resolution of recent sensors, object-based image analysis (OBIA) has gained importance for producing detailed land use maps. One of the main advantages of OBIA is that a variety of spectral, spatial and textural features can be extracted for the segmented image objects that are later utilized in classification. However, using a large number of features not only increases the required computational time, but also requires a large number of ground samples, which is unavailable in most cases. For these reasons, feature selection (FS) has become an important research topic for OBIA based classification studies. In this study, three filter-based FS algorithms namely, Chi square, information gain and ReliefF were applied to determine the most effective object features that ensure high separability among landscape features. For this purpose, importance degree (i.e. ranks) of 110 input object features were firstly estimated by the algorithms, and correlation-based merit function was then applied to determine optimum feature subset size. Multi- resolution segmentation algorithm was applied for segmenting a WorldView-2 image. Support vector machine, random forest and nearest neighbour classifiers were all utilized to classify segmented image objects using the selected object features. Results revealed that the FS algorithms were effective for selecting the most relevant features. Also, the classifiers produced the highest performances with 24 out of 110 features selected by the information gain (IG) algorithm. Partic- ularly, the support vector machine classifier produced the highest overall accuracy (92.00%) with 24 selected features determined by the IG algorithm. A significant improvement of about 4% was achieved by applying FS procedures that was found statistically significant in terms of Wilcoxon signed-ranks test. Keywords Object-based classification Á Feature selection Á Support vector machine Á Random forest Á Chi square Á Information gain Introduction Thematic maps produced through classification using remotely sensed imagery have been main data sources for many scientific applications at local to global scales. Cur- rent edge technology in recent sensors offers new oppor- tunities for many investigations in various research fields, such as environmental monitoring, urban land planning and change detection. These new sources of fine spatial resolution imagery will certainly increase the amount of information attainable for land cover and land use at local to global scales (Aplin et al. 1999). However, classification of such images is still a challenging task due to several factors affecting the success of a classification, such as the complexity of the landscape, high degree of within-class and between-class spectral variability and limitations of existing image-processing and classification approaches (Thomas et al. 2003; Lu and Weng 2007). With the increasing spatial resolution of satellite ima- gery in recent years, object-based image analysis (OBIA) has become a powerful approach producing superior per- formance in image analysis. Despite conventional pixel- based approach, object-based classification approach firstly forms homogenous image objects called segments from contagious pixels and then applies classification process on & Ismail Colkesen icolkesen@gtu.edu.tr Taskin Kavzoglu kavzoglu@gtu.edu.tr 1 Department of Geomatics Engineering, Gebze Technical University, 41400 Gebze, Kocaeli, Turkey 123 https://doi.org/10.1007/s12524-018-0807-xJournal of the Indian Society of Remote Sensing (August 2018) 46(8):1233–1242 Received: 13 February 2017 / Accepted: 22 January 2018 / Published online: 21July 2018