Edge detection is widely used in image processing, image recognition and computer vision. Tradition edge detection

methods have a lot of disadvantages. This paper presents a new edge detection algorithm base on Matching Pursuit

(MP). Our method has four steps. Firstly, the color image is converted to gray Image. Secondly, the residual part of

Matching Pursuit algorithm with the Wavelet Packet dictionary is extracted. Thirdly, the Gaussian Filter is applied to

remove the additional parts of the previous image and finally, the Ostu’s thresholding method converts image to

binary form. We compared our algorithm with the tradition methods like Prewitt, Sobel, Laplacian of Gaussian (LOG)

and Canny operators. Results show that our algorithm has high value of Performance parameters such as PR, PSNR

and FOM compared to other algorithms.