Grammars were used in computer science from the beginning for designing compilers for the first programming language. They were later used in image generation (for example, L-systems were used to generate beautiful pictures of plants). Therefore, using grammars for pattern recognition in images seems to be a natural research avenue.
This paper represents an overview of papers and research directions concerning the use of grammars for pattern recognition in images over the last decade. After a short introduction of the field and a discussion of older reviews, the paper continues with an explanation of how the reviewed papers where sampled.
The main body of the paper presents the review results. The results are stated quantitatively in the form of graphs and tables and are clustered according to similar techniques, objectives, or type of grammar used. A qualitative analysis section includes discussion of the different pattern recognition approaches in several short sections with references to the most important papers in the field.
Following the results, the authors discuss the advantages and limits of the use of grammars for pattern recognition and point to future research directions. A short conclusion closes the paper.
The paper is well written and gives a nice overview of the state of affairs with many useful references for in-depth study. This is a good starting point for anyone interested in techniques for pattern recognition in images. Due to the discussion of possible research directions, it is also good for anyone searching for research topics in this area.