Basic work principles of the SDK

Basic pipeline for fingerprint recognition consists of following steps:

1. The SDK decodes an image received as an input. See Image requirements here Requirements. See NeuroDactyl::Decoder and NeuroDactyl::Image.
2. The SDK detects fingerprint or fingerprints on an image. Result of the detection is used for following procedure of biometric template extraction. See NeuroDactyl::Detector and NeuroDactyl::Detection.
3. Detection result passes for extraction of biometric template – a compact descriptor, containing unique features of a particular fingerprint. See NeuroDactyl::Extractor and NeuroDactyl::Template
4. Fingerprint recognition is a process of matching biometric templates against each other.
   
   • Verification (1:1): matching of 2 templates against each other and getting recognition/similarity score
     
   • Identification (1:N): matching of one template against an enrollment database (Template DB) and getting TOP K of best matching results, for example, TOP-5, TOP-10, TOP-100, etc. best matching results.
     
   • SDK supports 1:N matching in batch mode (M:N), which means that you pass several templates at the same time for matching against the TemplateDB. M:N matching gives better recognition speed in terms of throughput comparing to 1:N matching. See NeuroDactyl::Matcher and NeuroDactyl::TemplateDB
     
5. All system configurations are set up in one Config. For test purposes you can use default config file. See Configuration.
   

See How to start with the SDK for detailed explanations of how to work with Neurodactyl fingerprint recognition SDK.
If you are an experienced user of biometric SDKs, see Code examples.