Image recognition is a mechanism to recognize an image that is not recognized by eyes, using certain method. This research was fingerprint recognition based on wavelet transforms and neural network. The aims of this research are to find the best wavelet and to know what the performance of this method is. Fingerprint recognition algorithms start from extracting an image to find image signature by choosing a little wavelet transforms coefficients that have the biggest magnitude value and neural network was used to select the best match (likeness) to original images in the collection.
The test were carried out in three kind of wavelets viz Coiflet 6, Daubechies 8, dan Symlet 8 and 5 types of query images (pure, blur, noise, pencil sketch, and edge) and each query image has 30 samples. Query's success rates were determined by using one percent threshold value times size of databases.
The result show that this method has good performance, which the average of success rate over 90% and need a little time query. The Symlet 6 can be considered to be the best wavelet for fingerprint image recognition, with success rate 96.36%. With respect to the elapsed query time, of about 0.11 second, the above method is sufficiently efficient for the database size of 1500 records.