| Eddie ward of Beijing science and technology development co., LTD., independent research and development of fingerprint identification algorithm "IDworld v5.0" using both unique and fast fingerprint image processing technique and high speed matching algorithm.Therefore, can be used in database systems with large capacity, can also be applied in the low frequency of CPU is an embedded system and ensure the high degree of recognition performance. "IDworld v5.0" algorithm has the following features. Identification of a highly accurate rates In the "IDworld v5.0" algorithm, a new developed has strong recognition fingerprint ribbed line flow characteristics, to complete a new project, combined with the feature point information using to ensure the stability and high accuracy. Use "IDworld v5.0" algorithm, can in their wrinkles or scars much fingerprint image, sweat too much fingerprint image and dry fingerprint image to ensure good recognition performance. 5 * 1011 times than been conducted, the results are as follows. - to recognize false rate (FAR) : less than 0.00001% Boomed - rejected (FRR) : less than 0.1% High speed matching rate Is highly than rate "IDworld v5.0" algorithm is one of the main characteristics. "IDworld v5.0" algorithm, in addition to relying on many engine (using) the relationship between the feature points of high technology, has adopted the rib based on fingerprint line flow characteristics of the continuous indexing methods.As a result, the algorithm has large capacity of the database system also ensures fast alignment. Feature extraction is both accurate and fast "IDworld v5.0" algorithm, because of the unique image processing technology, so it is very accurately extract the feature, can greatly shorten the extraction time. To extract the characteristic time for 5 to 10 milliseconds (Pentium 4,2.8 GHz conditions). "IDworld v5.0" algorithm based on new panoramic prediction method and residual like features research field of fingerprint extraction method.For dry, therefore, the fingerprint can quickly and accurately extract the fingerprint area. Registration data occupies little capacity. Use "IDworld v5.0" algorithm, using many identification characteristics, not only can greatly improve the recognition performance, can also reduce the size of the registration data to 496 bytes.This size is the other engine. Practical intelligent learning function The company successfully developed a method of feature added this method ensure use to identify the new characteristics of information in the process of constantly update the registered fingerprint information, so that the poor quality of the fingerprint characteristic also with the use of constantly improve and ensure the stable and correct recognition. Alignment model had nothing to do with rotation and translation. "IDworld v5.0" algorithm adopted by the coordinate transformation based on feature is not affected by relativity and direction of the axis symmetry, the impact of rotation and translation.Therefore, even if any position, put the finger on the fingerprint head can also be stable. Comparing the distortion image stable. "IDworld v5.0" algorithm USES the nonlinear distortion image matching method is one of the TPS method, so that the distortion of fingerprint image is stable and accurate than. Fingerprint acquisition head generality Use "IDworld v5.0" algorithm, as long as the guarantee of more than 400 dpi image resolution, to use any collection head (optical, static type, pressure type, thermal type, etc.) are able to do a complete fingerprint images collected ribbed line recovery. Flexible corresponding hardware conditions. "IDworld v5.0" algorithm for low memory requirement, the fingerprint processing speed is fast.Therefore, as long as it is in the 64 KB of RAM, 96 KB of coding with Flash and 50 MHZ CPU conditions, can ensure a high degree of recognition performance. |