Performance comparison between RGB and HSI linear stretching for tuberculosis bacilli detection in Ziehl-Neelsen tissue slide images

Abstract

Ziehl-Neelsen staining method is commonly used for diagnosis of tuberculosis patients. Sputum or tissues samples are stained using this method before microscopic examination. However, manual screening is time consuming and tedious process. Therefore, an automated image processing and analysis system for improving the manual screening of samples is desirable. Unfortumately, the quality of images being observed is influenced by some factors such as illumination, staining and noise problem. Image enhancement is a necessary part of image processing technique used to improve the quality of images being captured. This paper investigates the performance of linear stretching based on RGB and HSI colour spaces. In RGB colour space, linear stretching of each RGB component is carried out, while in HSI colour space, linear stretching is performed on S and I components. Both RGB and HSI linear stretchign are applies to enhance the Ziehl-Neelsen tissue slide images that have been analysed under light microscope and capture using digital camera. Although both techniques are quite simple, the results indicate that these methods may have potential to be used in enhancing Ziehl-Neelsen tissue slide images. However, overall analyses indicates that linear stretching based on HSI colour spaces performed better enhancement compares to RGB based linear stretching.