Image Reconstruction Validation for CMOS Linear Image Sensor Based Tomography System

Abstract

Nowadays, the fast-moving industries need a low cost, accurate, low power consumption, non-invasive and safe method for monitoring and tracking of the conveying and manufacturing process. There is many tomography based systems available in the industry, however, with limitations. This project aims to solve industrial problems of monitoring objects inside a transparent conveying pipe and determine the characteristic of the objects such as size, quantity and position. This is done through the image reconstruction of data acquired from Complementary metal-oxide-semiconductor (CMOS) linear image sensor based on optical tomography system. Data from four projections of laser are used to avoid the aliasing problem that may occur due to fewer projections. The image can be reconstructed by using linear back projection technique. The image will undergo image processing to enhance the image for better visualization of the object. Theoretical and experimental image are compared to validate the image reconstruction. As a result, the position, size and number of a symmetrical and solid object can be determined accurately from the reconstructed image. For further improvement, the number of projection may be increased.