A Feature Extraction Method for Recognition of Petechia Dot in Tongue Image

Petechia Dot ranks itself among the most significant pathological features utilized in the practice of the Tongue Diagnosis because it can reveal the existence of tiny hemorrhage in some organ inside the body. In this paper, a feature extraction method aiming for recognition of the Petechia Dot in tongue image is proposed employing graylevel-height based thresholding, Top-hat high-pass filtering, and Gaussian histogram specification. Graylevel contrast-stretch and morphological filters are designed as preprocessing method to make Petechia Dot more apparent. Finally, algorithm erasing falsely selected pixel is presented to avoid misjudgment with help of region growing and Hit-or-Miss transform. Experiments to test correctness of these methods are carried out, and result is proved to be satisfactory.