A passive, millimeter wave (MMW) and terahertz (THz) dual-band imaging system composed of 94 and 250 GHz single-element detectors was used to investigate preprocessing and fusion algorithms for dual-band images. Subsequently, an MMW and THz image preprocessing and fusion integrated algorithm (MMW-THz IPFIA) was developed. In the algorithm, a block-matching and three-dimensional filtering denoising algorithm is employed to filter noise, an adaptive histogram equalization algorithm to enhance images, an intensity-based registration algorithm to register images, and a wavelet-based image fusion algorithm to fuse the preprocessed images. The performance of the algorithm was analyzed by calculating the SNR and information entropy of the actual images. This algorithm effectively reduces the image noise and improves the level of detail in the images. Since the algorithm improves the performance of the investigated imaging system, it should support practical technological applications. Because the system responds to blackbody radiation, its improvement is quantified herein using the static performance parameter commonly employed for thermal imaging systems, namely, the minimum detectable temperature difference (MDTD). An experiment was conducted in which the system’s MDTD was measured before and after applying the MMW-THz IPFIA, verifying the improved performance that can be realized through its application.
Compared with microwave, THz has higher resolution, and compared with infrared, THz has better penetrability. Human body radiate THz also, its photon energy is low, it is harmless to human body. So THz has great potential applications in the body searching system. Dual-band images may contain different information for the same scene, so THz dual-band imaging have been a significant research subject of THz technology.<p> </p>Base on the dual-band THz passive imaging system which is composed of a 94GHz and a 250GHz cell detector, this paper researched the preprocessing and fusion algorithm for THz dual-band images. Firstly, THz images have such problems: large noise, low SNR, low contrast, low details. Secondly, the stability problem of the optical mechanical scanning system makes the images less repetitive, obvious stripes and low definition. Aiming at these situations, this paper used the BM3D de-noising algorithm to filter noise and correct the scanning problem. Furthermore, translation, rotation and scaling exist between the two images, after registered by the intensity-base registration algorithm, and enhanced by the adaptive histogram equalization algorithm, the images are fused by image fusion algorithm based on wavelet. This effectively reduced the image noise, scan distortion and matching error, improved the details, enhanced the contrast. It is helpful to improve the detection efficiency of hidden objects too. Method in this paper has a substantial effect for improving the dual-band THz passive imaging system’s performance and promoting technology practical.