Fast and comprehensive FPGA based BLOB analysis with the Hybrid BLOB concept

In this contribution we show our approach for a feature rich and high speed BLOB analysis on FPGAs. For the Hybrid-BLOB concept we use a combination of a single-pass BLOB analysis and a double-pass labeling algorithm. We use Basler’s VisualApplets for the implementation of the concept on their microEnable 5 frame grabbers. We achieve the extraction of the gray value data of the BLOBs at factor 14 higher frame rates compared to the naive labeling of the complete image. This is achieved by limiting the maximum BLOB size to 128 × 128 px, which speeds up the double-pass labeling algorithm. Our con-cept is targeted at low latency and high throughput demanding applications where BLOBs are small, like sensor based sorting or surface inspection.