A Methodology to Align Existing Vector Layers Used for Roads with High Resolution Satellite Images

Abstract

The latest products based on geographical information are collected spatial data with higher accuracy and improved processing techniques. With the frequent growth of geospatial information processing techniques the capability of combining various sets of geospatial data into a single composite vector dataset is becoming one of slightly increasing dominant issues. However it‟s a challenging task to automatically conflate vector data with its geo referenced imagery. Since geospatial data obtained from various data sources they may have different projections, different accuracy levels thus resulting in various positional inconsistencies. Though we apply same coordination and projection to vector to imagery mapping, still we can see a misalignment of them. In this research work, we describe an automatic novel geospatial data fusion approach to conflate the vector road layer data with its corresponding imagery road layer. Our general approach to vector to raster conflation examines the problem through many phases. In the initial stage intersection points in the vector layer are identified using Hough transformation based technique. Afterward to find their corresponding intersection points in the raster layer we use localized image processing technique based on Bayes classification. At the end of the process Snake algorithm is applied to optimize the conflation confusion of two layers. Finally the proposed methodology has successfully addressed the challenge of vector layer to imagery conflation.