Improved Sub-Pixel Mapping Method Coupling Spatial Dependence With Directivity and Connectivity

Abstract

Accurate land cover mapping by using coarse resolution imageries has been an attractive research topic. Sub-pixel mapping has been proven efficient for allocating sub-pixels within a mixed pixel. The most likely distribution can be determined on the condition of maximized spatial dependence. However, linear land cover like roads and rivers cannot be predicted efficiently because of weaker spatial dependence between and within mixed pixels. To obtain more accurate classification at the sub-pixel scale, an improved sub-pixel mapping method by combining spatial dependence with directivity and connectivity of linear land cover was proposed. Central line of linear land cover was extracted from fraction images to provide site-specific information. Discriminated allocation targets were accordingly designed: both connectivity and directivity were considered as important auxiliary information for allocating linear land cover, whereas only maximized spatial dependence is required for other classes. Then, simulated annealing arithmetic (SAA) was applied to optimize sub-pixel allocation. The method was evaluated visually and quantitatively with the accuracy indices. Compared with the model that considers only spatial dependence, SPM HIIPD method, attraction model and hard classifier (MLC), the improved method can increase classification accuracy at the sub-pixel scale with both simulated imageries and partial SPOT remotely sensed imagery.Index Terms-Directivity and connectivity, simulated annealing arithmetic (SAA), spatial dependence, sub-pixel mapping.

Principles and methods of scaling geospatial Earth science data

“…Moreover, neither spatial autocorrelation-based nor spatial pattern-based SRM methods effectively preserve the structure of linear patterns. Anisotropic models (Thornton et al, 2007), linear subpixel mapping agents (Xu et al, 2014), central line control (Ai et al, 2014), and linear templates (Ge et al, 2016a) (Nigussie et al, 2011;Wang et al, 2006;Zhang et al, 2008), support vector regression (Zhang et al, 2014c), or the patch-pair learning-database (Ling et al, 2016).…”

“…These methods ignore the local details or the structure of coarse patch features. Downscaling categories Modeling spatial distribution patterns of geographical featuresTatem et al, 2002;Atkinson, 2004;Thornton et al, 2007;Xu et al, 2014;Ai et al, 2014;Ge et al, 2016a The features of categorical variables are divided into areal pattern, point pattern and linear pattern, which are described by building different models. The accurate division of the three patterns is key and prerequisite for SRM.…”

Foreword to the Special Issue on Pattern Recognition in Remote Sensing

“…In [29], subpixel mapping is addressed, which has been proven efficient for allocating subpixels within a mixed pixel. To obtain more accurate mapping at the subpixel scale, an improved method combining spatial dependence with directivity and connectivity of linear land covers is proposed, and simulated annealing arithmetic (SAA) is applied to optimize subpixel allocation.…”

Spatial–Temporal Super-Resolution Land Cover Mapping With a Local Spatial–Temporal Dependence Model

“…The a-priori information in STD_STSRM includes the a-priori spatial information that is used to predict the land cover spatial patterns at the fine resolution pixel scale and a-priori temporal information that is used to model the temporal transitions between the class labels in the predicted and the input pre-or post-dated land cover maps. The a-priori spatial models have been studied in SRM researches including the spatial dependent model [20][21][22][23], the direct mapping model [24], the geostatistical model [25,26], the multi-point simulation based model [27], the learning based model [28,29], the adaptive model [30], and the linear spatial distribution model [31,32]. In these models, [20][21][22][23][24] are suitable for predicting spatial patterns of patches that are larger than the coarse resolution pixel, [25][26][27] are suitable for predicting spatial patterns of patches that are smaller than the coarse resolution pixel, [31,32] are suitable for linear patch, and [28][29][30] are suitable for patches with different spatial patterns, respectively.…”

“…The a-priori spatial models have been studied in SRM researches including the spatial dependent model [20][21][22][23], the direct mapping model [24], the geostatistical model [25,26], the multi-point simulation based model [27], the learning based model [28,29], the adaptive model [30], and the linear spatial distribution model [31,32]. In these models, [20][21][22][23][24] are suitable for predicting spatial patterns of patches that are larger than the coarse resolution pixel, [25][26][27] are suitable for predicting spatial patterns of patches that are smaller than the coarse resolution pixel, [31,32] are suitable for linear patch, and [28][29][30] are suitable for patches with different spatial patterns, respectively. These a-priori spatial models used in SRM can be directly applied in STD_STSRM.…”