Image Segmentation Applied To Nail Diagnosis Of TCM

The observation diagnosis of Traditional Chinese Medicine (TCM) is used by the doctor to infer whether the patient is sick or not and what kind of disease the patient is suffering from according to the obtained the symptoms by observing relevant parts of the tissues and organs of the patient. In order to inherit and carry Traditional Chinese Medicine forward, this paper attempts to use image processing technology of computer to achieve nail diagnosis. The nail diagnosis of this paper is combined with advanced science and technology, which will be helpful to traditional Chinese medicine. The nail diagnosis based on image processing technology includes nail image acquisition, image segmentation, image feature extraction, disease detection and recognition algorithm. Due to the lack of a standard image database used to infer diseases and the time, this paper is mainly about preliminary research of the nail diagnosis based on image processing technology, including the study of the pixels of the nail image, the nail image segmentation algorithm, the improvement and the accomplishment of the algorithm. Specifically, the study is done as follows:1. The paper studies the features of the original image pixels collected under the specific external environment in RGB color space and YCbCr color space and analyzes the pixels of the nail part and the skin part of the original image, including the value and the threshold of these pixels. The values of the pixels of the nail image and the skin image are calculated in YCbCr color space by the means of sampling, and the effects of light conditions on the nail image are explored.2. The original image is segmented by the means of threshold-based segmentation algorithm, Edge detection method, k-means clustering algorithm and watershed algorithm. The results are compared and the watershed algorithm works best of the four methods. Because the watershed algorithm needs to mark the nails image segmentation by manual assistance, the watershed algorithm based on k-means is used as the final nail image segmentation method. After the analysis of the advantages and disadvantages of the algorithm, an improved watershed algorithm based on k-means is proposed for the nail image segmentation. The improved watershed algorithm based on k-means will automatically and perfectly mark the target image and use the marked image to achieve the nail image segmentation under the specific external environment.3. The hardware and software environment is built on the Zedboard platform, including the design of HDMI, Linux system porting, OpenCV and QT porting. The detailed code which can exactly show how the improved watershed algorithm works is accomplished and tested on the platform after building the hardware and software environment. According to the outcome, the program is improved.The most difficult problem of the nail diagnosis based on image processing technology is the lack of standard human fingernail image database. As the key to the next research work, the image database will be used to build models of diseases and study the disease recognition algorithm.