Performance evaluation of the various edge detectors and filters for the noisy IR images

Edge detection is an important process of image processing and it is the foundation of pattern recognition, which will affect the later processes. Especially in a noisy image, the edge detection is more important because noise is a common phenomenon in an image. This paper deeply studies the edge detection methods for noisy IR images and also studies the performance of the filters involved in removing the noise. Four basic edge detection operators such as Sobel, Prewitt, Log and Canny were selected and a comparison is done to check the quality of the edge detectors with noisy IR images. The study shows that the Sobel edge detector is best in detecting the edges when compared to other edge detectors. Then, the filters such as Median, Lee and Kuan were used to remove the noise in IR image. After applying all the filters the result shows that Median filter works good among the three filters. The quality of the edge detectors and the filter is calculated using the parameters PSNR and RMSE.