Lossless image compression by block matching on a mesh of trees

Summary form only given. In this paper, we showed a work-optimal parallel algorithm using the rectangle greedy matching technique requiring O (log M log n) time on the PRAM EREW model. We showed how algorithm is implemented on a mesh of trees still with optimal parallel work and in O (log M log n) time. Differently from arrays and trees, meshes of trees have both small diameter and large bisection width, which makes them as fast as hypercubic networks but simpler to build. In our case, we can even run the PRAM algorithm on the mesh of trees without slowing it down and without increasing the number of processors