Normalization in Unsupervised Segmentation Parameter Optimization: A Solution Based on Local Regression Trend Analysis

Stefanos Georganos,Brian Johnson,Eléonore Wolff,Sabine Vanhuysse,Tais Grippa,Moritz Lennert

doi:10.3390/rs10020222

Stefanos Georganos, Brian Johnson + Show 4 more

Open Access

https://doi.org/10.3390/rs10020222

Copy DOI

Abstract

In object-based image analysis (OBIA), the appropriate parametrization of segmentation algorithms is crucial for obtaining satisfactory image classification results. One of the ways this can be done is by unsupervised segmentation parameter optimization (USPO). A popular USPO method does this through the optimization of a “global score” (GS), which minimizes intrasegment heterogeneity and maximizes intersegment heterogeneity. However, the calculated GS values are sensitive to the minimum and maximum ranges of the candidate segmentations. Previous research proposed the use of fixed minimum/maximum threshold values for the intrasegment/intersegment heterogeneity measures to deal with the sensitivity of user-defined ranges, but the performance of this approach has not been investigated in detail. In the context of a remote sensing very-high-resolution urban application, we show the limitations of the fixed threshold approach, both in a theoretical and applied manner, and instead propose a novel solution to identify the range of candidate segmentations using local regression trend analysis. We found that the proposed approach showed significant improvements over the use of fixed minimum/maximum values, is less subjective than user-defined threshold values and, thus, can be of merit for a fully automated procedure and big data applications.

Highlights

IntroductionFor high spatial resolution Remote Sensing (RS) images, it is often beneficial to perform image processing (e.g., image segmentation or image filtering) prior to image classification
For high spatial resolution Remote Sensing (RS) images, it is often beneficial to perform image processing prior to image classification
Global Moran’s I (MI) and area-weighted variance (WV) values are often combined for this purpose, and an “optimal” segmentation is identified based on the combined result; i.e., the “Global Score” (GS) values of the candidate segmentations

Summary

Introduction

For high spatial resolution Remote Sensing (RS) images, it is often beneficial to perform image processing (e.g., image segmentation or image filtering) prior to image classification. Object-based image analysis (OBIA) has been increasing in popularity in the past years, with many studies reporting advantages over a pixel-based approach for RS data of various scales and resolutions, and a clear-cut benefit for very high-resolution (VHR) imagery [2,3,4,5]. As the segmentation step is of significant importance with respect to classification accuracy, appropriate parametrization of the segmentation algorithm is required [7,8]. This parametrization is typically done using supervised, semi supervised or unsupervised techniques [9,10,11,12,13,14,15]

Methods

Results

Discussion

Conclusion