Color Image Enhancement Techniques for Endoscopic images

Abstract

Modern endoscopes play an important role in diagnosing various gastrointestinal (GI) tract related diseases. Although clinical findings of modern endoscopic imaging techniques are encouraging, there still remains much room for improvement of image quality. Of greatest concern, endoscopic images suffer from various degradations, such as specular highlights, non-uniform brightness and poor contrast. As a result, gastroenterologists often face difficulty in successfully identifying the subtle features, such as mucosal surface and structures, pit patterns, size and pattern of micro-vessels, tissue and vascular characteristics, superficial layer of mucosal and abnormal growths in endoscopic images. The improved visual quality of images can provide better diagnosis. This paper presents two proposed post-processing techniques for enhancing the subtle features of endoscopic images. The first proposed technique is named as endoscopic image enhancement based on adaptive sigmoid function and space-variant color reproduction (ASSVCR). It is achieved in two stages: image enhancement at gray level followed by color reproduction with the help of space variant chrominance mapping. Image enhancement is achieved by performing adaptive sigmoid function and uniform distribution of sigmoid pixels. Then color reproduction is used to generate new chrominance components. The second proposed technique is named as tri-scan. It is achieved in three stages: (1) Tissue and surface enhancement: a modified linear unsharp masking is used to sharpen the surface and edges of tissue and vascular characteristics, (2) Mucosa layer enhancement: an adaptive sigmoid function similar to the ASSVCR technique is employed on the R plane of the image to highlight the superficial layers of mucosa, (3) Color tone enhancement: the pixels are uniformly distributed to create a different color effect to highlight mucosa structures, superficial layers of mucosa and tissue characteristics. Both techniques are compared with other related works. Several performance metrics like focus value, statistic of visual representation, measurement of uniform distribution, color similarity test, color enhancement factor (CEF) and time complexity are used to assess the performance. The results showed improved performance compared to similar existing methods. In the post-processed images, we have observed that the ASSVCR can enhance and highlight pit patterns, tissue and vascular characteristics, mucosa structures and abnormal growths. It cannot highlight size and pattern of micro-vessels, and superficial layer of mucosa. In contrast, tri-scan can enhance and highlight all above mentioned features of endoscopic images.