Compression of Pulsed Terahertz Image Using Discrete Wavelet Transform

Abstract

We present the result of data compression using discrete wavelet transform (DWT) in terahertz pulsed imaging (TPI). A test object composed of dielectric materials is imaged by a fast oscillating delay line in reflection geometry. In general, because the size of the original image is from several megabytes (MB) to tens of MB, an adequate compression algorithm that prevents the loss of relevant information is indispensable. We analyze the image by multilevel DWT and obtain the wavelet coefficients of the levels. The images reconstructed from some of the coefficients are compared with the original image. The reconstructed pulse signals and their frequency-domain spectrums for each pixel are compared with the original signals to obtain the optimum decomposition level. The image could be compressed below 15% of its original size without loss of image quality and significant errors in the time and frequency domains.