Hardware Acceleration of BLOB Detection for Image Processing

This paper presents the implementation and evaluation of a computer vision task on a Field Programmable Gate Array (FPGA). As an experimental approach for an application-specific image-processing problem, it provides results about gained performance and precision compared with similar solutions on General Purpose Processor (GPP) architectures. The problem of detecting Binary Large OBjects (BLOBs) in a continuous video stream and computation of their center points has been addressed. Most existing solutions are realized on GPP platforms, where resolution of image material and sequential processing define the performance barrier. FPGA based approaches perform implemented algorithms as fast as hardware circuits and in addition offer parallelization abilities. The evaluation compares precision and performance gain against similar approaches on GPP platforms. The paper discusses different concepts for BLOB detection and shows the implementation of one common method for BLOB detection, including design problems and performance evaluation.