Improving the Quality of Multi-spectral Transmission Image Based on Modulation-frame Accumulation and Local Linear Embedding Algorithm

Abstract

In biological tissue, optical transmission imaging is blurred due to absorption and scattering effects, which makes it difficult to detect heterogeneities on multi-spectral transmission image. In this paper, the inter-combination of modulation-frame accumulation and local linear embedding (LLE) super-resolution (SR) image reconstruction algorithm is introduced into multi-spectral transmission images for the first time to improve the quality of image. Firstly, the acquisition experiment of signal modulated phantom multi-spectral transmission image is designed. Secondly, frame accumulation processing is performed on the obtained multi-spectral transmission image. Meanwhile, the four-wavelength images with the same concentration ratio are recombined every three wavelengths to obtain true-color images, and the fusion principle is carried out in sequence of blue, green, red and near-infrared light. Then, multi-spectral true-color images are processed by LLE-SR model. Finally, the improvement of image quality is verified by full reference image quality assessment (FR-IQA). The experimental results show that frame accumulation, original-LLE and frame accumulation-LLE models proposed in this paper can improve the quality of multi-spectral transmission images. Among them, the original-LLE model in the FR-IQA index shows that the use of local linear correlation to restore the high precision of multi-spectral transmission image is more effective. In conclusion, based on the inter-combination of the above different image processing methods, this paper achieves the acquisition of high-quality multi-spectral transmission images, which will further improve the possibility of the application of optical transmission imaging technology in clinical screening of breast tumors.