Performance evaluation of various image features detectors and descriptors used in the development of panorama from real time video files

Abstract

Image mosaicing is a method where two or more pictures of the same image can be combined into a big picture and a high resolution panorama created. It is helpful for constructing a bigger picture with numerous overlapping pictures of the same scene. The image mosaic development is the union of two pictures. The significance of image mosaicing in the sector of computer vision, medical imaging, satellite data, army automatic target recognition can be seen. Picture stitching can be performed from a broad angle video taken from left to right to develop a wide-scale panorama to obtain a high-resolution picture. This research paper includes valuable content which will be very helpful for creating significant choices in vision-based apps and is intended primarily to establish a benchmark for scientists, regardless of their specific fields. In this paper it has been seen that distinct algorithms perform differently in terms of time complexity and image quality. We have looked at a variety of feature detectors and descriptors such as SIFT-SIFT, SURF-SURF, STAR-BRIEF and ORB-ORB for the development of video file panoramic images. We have noted that SIFT provides excellent outcomes, giving the image the largest amount of key points identified at the cost of computational time and SURF, ORB, has fewer key points obtained, where it has been seen that ORB is the simplest of the above algorithms, but produces no good performance quality image outcomes. A good compromise can be achieved with SURF. Depending on the application, the metric for image feature extraction would change. In addition, the speed of each algorithm is also recorded. This systemic analysis suggests many characteristics of the stitching of images.