CONNECTED COMPONENT LABELING FOR BINARY IMAGES.

Connected Component Labeling is one of the very important aspects of Image Processing and Computer Vision. Connected Components refers to set of pixels having same value connected to each other in way that there exists a path between every two pixel of the connected component set.This project proposes 3 different algorithms related to different perspectives to solve Connected Component Labelling in Binary Image. These 3 perspectives are: Scaling, N-dimensions, Parallel. Scaling: This algorithm reduces the resolution of image and then CCA is performed on the low resolution image. After this, label matrix is expanded to high resolution. Then accretion is done to resolve irregular labels. Basic idea behind this approach is that lesser the number of pixels fast is the execution of CCA/L algorithm. N-dimensions: Algorithm can handle any n-dimensional image, so it works for 1, 2...n-dimensions. This important because we can have higher dimensional images like 20-D or more in near future. Parallel: Image will be processed simultaneously on separate processors and results will be merged and then sorted to produce a single label matrix corresponding to original image. This consumes less memory and less execution time. It provides results fast for images of varying size and densities. All the approaches seem to give good performance. They produce accurate results and are efficient in terms of memory consumption and speed.