GPU实区块CCL,This report is a documentation of the workings of an NVIDIA® CUDA (Compute Unified Device Architecture) based application which implements a parallelized version of the Connected Component Labelling algorithm on an NVIDIA® Graphics Processing Unit (GPU) alongwith various other image smoothening algorithms to achieve significant speedups. This includes details about running times and speedup gains achieved by parallelizing traditional serial CCL (Connected Component Labelling) algorithms.