Enhanced SIFT for Image Registration of VHR Images

Abstract

This paper mainly focuses on effectively improving the working of adaptive contrast threshold of the SIFT detector. Putting aside the fact that SIFT detector based calculations have a huge place when it comes to its usage in the remote detecting panchromatic picture coordinating, yet some small fixations in contrast threshold can accomplish outstanding changes in the image matching quality. The ACT-SIFT technique calculates the contrast thresholds for the source and target image. Therefore, a goal can be achieved through limiting some exceptional measures. Taking an early step, the required key points are first separated from the scale space of SIFT method by using the differentiation edge of the two pictures. Later, complexity limit is continuously changed at regular intervals of time or in other words iteratively changing the threshold values for the targeted and reference pictures. The right matching pairs are also created with the help of the matches obtained through the SIFT descriptor. The results obtained by using this technique increased the matching pairs in the remote sensing image correspondence, which takes out the key points by checking the data of the two pictures at the same time.