Abstract: To refine image characters can help to highlight the shape features of the character and reduce
the amount of redundant information, which has important applications in the field of character
recognition. After analyzing and studying the traditional thinning algorithm, aiming at the distortion and
incomplete phenomenon of traditional refinement, this paper proposes a refinement method for the
characters of international phonetic symbols. The algorithm makes it possible to refine the width of the
international phonetic image character to a pixel width by marking the edge of the text area and judging
whether the marked point satisfies the removable condition and then gradually removing the edge pixel.
Experiment on the international phonetic alphabet character image shows that the algorithm proposed in
this paper can accurately refine the international phonetic alphabet character image, simple and efficiently.

Key words: image, international phonetic alphabet, character, thinning