We have presented efficient data parallel algorithms for texture analysis and synthesis based on Gibbs or Markov random field models. A complete software package running on the Connection Machine model CM-2 and the Connection Machine model CM-5 implementing these algorithms is available for distribution to interested parties. The experimental data strongly support the analysis concerning the scalability of our algorithms. The same type of algorithms can be used to handle other image processing algorithms such as image estimation ([8], [9]), texture segmentation ([5]), and integration of early vision modules. We are currently examining several of these extensions.
The parameters for the 
image of tree bark texture in
Figure 5 are given in Table 2.