Exemplar-based texture synthesis is a process of generating perceptually equivalent textures with the exemplar. The present work proposes a novel patch-based synthesis algorithm for synthesizing new textures that employs the powerful concept of gray-level co-occurrence matrix coupled with restricted cross-correlation. Furthermore, a simple and peculiar blending mechanism has been devised which avoids the necessity of retracing the path after ascertaining the minimum cut within the overlap region between the two neighboring patches. The method has been tested and executed for the samples derived from Brodatz album, the widely acceptable benchmark dataset for texture processing. The results are found to be comparable to Efros and Freeman for stochastic texture while outperforms the Efros and Freeman algorithm for semistructured texture.