Change Vector Analysis using Enhanced PCA and Inverse Triangular Function-based Thresholding

摘要

Change vector analysis is a very sophisticated method to evaluate land-use/land-cover changes meaningfully. By making proper choice of input data in the form of bands (for instance, red, NIR etc) or features (for instance, greenness, brightness, wetness etc), information about both the magnitude as well as the typeature of changes can be extracted. However, improper selection of thresholds is always a hindrance to a good change detection algorithm. The paper has proposed an improved technique to select threshold appropriately by means of principal component difference and inverse triangular function. The changes have been represented using class-based circular wheel representation. Results have been shown to further testify the performance of proposed algorithm.