Comparison of various edge detection filters for ANPR

Automatic Number Plate Recognition (ANPR) can be identified as a technology which has been developed mainly based on Image Processing methodologies. It is being widely used in identifying vehicles in applications such as red-light enforcement, over speeding, bus lane control, motorway road tolling, border control and access/parking control. However axioms of ANPR rest on Image processing approaches such as binary conversion, morphological functions, filtering and edge detection algorithms. By researching on the solutions which have been suggested through numerous approaches to implement ANPR, this paper will focus on edge detection methodologies. A comparison is made between the edge detection filters Sobel, Canny, Gabor and Log-Gabor, using a variation of preprocessing techniques and using template matching as character recognition method. The algorithms are compared for accuracy as well as processing times. It is observed that as an all-round algorithm, Log-Gabor outperforms others. In terms of processing time, it is second only to Sobel. It has not been used in ANPR and hence has been identified as a suitable candidate for ANPR.