Analysis of edge detection algorithms for feature extraction in satellite images

In field of image processing and pattern recognition, the use of edges as a feature is significant for feature extraction owing to its simplicity and accuracy. Its areas of application vary from object recognition to satellite based terrain recognition. There are many edge detection techniques like the canny edge detector and Sobel edge detector etc. However, the quality of an efficient algorithm depends on its capability to generate well localized edges of real images. Noise is inherent in all real images. To reduce its effect, various smoothing low pass filters are used prior to edge detection. This work, compares edge detection based on bilateral filtering with canny edge detection technique for satellite Images. Bilateral filtering based edge detection not only generates well localized edges but also simultaneously reduces considerable noise from real life images. The results show that the bilateral filtering based edge detection provide better edge maps than other comparable techniques.