Design of a digital system that performs autonomous license plate feature extraction from an image using morphological transformations has been investigated. An algorithm has been developed that locates the license plate in a three step process. The process includes operations on the difference image obtained from dilation and erosion operations. The algorithm has been found to have an overall success ratio of 86.5% in tests performed using Matlab on 2 databases that total 564 selected images of cars. A VHDL based design that incorporates the algorithm has been developed and post-map simulation of the VHDL code successfully completed. In simulations, the design was found to perform at a maximum clock frequency of approximately 129 MHz with a transformation kernel size of 3. The clock frequency is reduced with increasing kernel size.