Image characteristic extraction method based on wavelet packet and mathematical morphology
Zhang Liguo, Yang Jin, Li Jing, Ren Xiaoli, Shangguan Hanlu(Institute of Electrical Engineering, The Key Lab of Measurement Technology and Instrumentation ofHebei Province, Yanshan University, Qinhuangdao 066004, China)
This paper proposes an image characteristic extraction method based on wavelet packet and mathematical morphology, and applies the method to the character extraction and edge detection of the images of license plate and etc. Wavelet packet is used to decompose the images and reconstruct their approximate part; edge detection of the images is carried out using morphological basic operations, such as dilation, erosion and morphological gradient; and the void of the plate characters is filled using morphological reconstruction for the identification of the characters of the license plate. Simulation experiments show that compared with edge detection operator of "edge function", the proposed method can extract the characteristic of the images better; and the extracted edge is obviously clear and the void of the characters is filled completely.