查詢結果分析
來源資料
相關文獻
- 中醫望診系統:彩色舌診影像系統之研發
- 肺腫瘤CT影像的分析技術及效能評估
- 不同影像處理演算法於梗塞性中風磁振影像的分析技術
- 磁振擴散加權照影術在梗塞性中風區域的分析技術
- Form Segmentation and Component Classification for Clinic Document Image Analysis
- Visualization and 3D Reconstruction for Ultrasound Images of Carotid Arteries with Plaques
- 進階雙預測器DPCM影像傳輸系統
- 類神經網路與貝氏分類法應用於影像分割之比較研究
- 多波段影像特徵分析於灘線萃取之研究
- 結合塔像與紋理分類之地物萃取法
頁籤選單縮合
題 名 | 中醫望診系統:彩色舌診影像系統之研發=Development of Color Tongue Viewing Diagnosis System for Chinese Medicine |
---|---|
作 者 | 蘇振隆; | 書刊名 | 中醫藥年報 |
卷 期 | 16:1 1998.05[民87.05] |
頁 次 | 頁427-521 |
分類號 | 413.241 |
關鍵詞 | 影像校正; 邊緣檢測; 影像分割; 舌影像特徵辨識; Image calibration; Edge detection; Image segmentation; Tongue image character recognition; |
語 文 | 中文(Chinese) |
中文摘要 | 舌診是中醫望診的重要項目,中醫師在診斷病情時,主要觀察舌的 神、色、形、態。如何利用電腦來進行舌診的定量分析,是近年來的趨勢。 本計畫的目的,是利用彩色影像處理的技術,對舌診所要觀察的特徵如舌 色、舌形、苔色做辨識,並藉由軟體與硬體的整合建立一個舌診定量分析 系統。 系統中包含影像擷取、環境校正與舌特徵分析等流程。在利用彩色攝 影機取得舌影像後,進行前處理工作。其工作包括整體的校正與RGB標準 的校正,得到的結果,來補償不同環境下取像可能造成的偏差。在舌形的 分析上,利用邊緣檢測的的方法,得到舌影像的邊緣,來算出舌長與舌寬, 並算出舌面積。此外,利用合成舌影像算舌之厚薄。在特徵的分析上,利 用15個健康人的舌影像進行分割後,建立辨識法。研究中,除了對整個 舌影像作分析與辨識;也依據中醫舌診的原理,將舌影像分成四部份去分 析舌的特徵分佈。 在系統完成後,由五十個人健康人的舌影像特徵的分析,歸納出健康 人舌象分佈的特點及舌質與舌苔變化的規律性。在整體的舌特徵分佈的百 分比結果上,整個舌影像舌色分佈在50%以上,其中淡紅色佔40%以上, 紅色與暗紅色則分佈在10%以下。在苔色的分佈上,大多數人之白苔為20% 以上,黃苔則少於白苔。此外在分辨黃苔與白苔時,約僅有0.5%無法確定, 其效果良好。舌寬在4到5公分,舌長在4到6公分,而舌面積則分佈在 15到25平方公分,是判斷舌形的依據。 在各部份舌特徵分佈的百分比上,在舌中部份舌色與苔色的分佈並沒 有很大的差異,左右的部份則是舌色分佈約佔40%,而在舌尖舌苔分佈是 最少的在20%之內。舌色的分佈上,舌尖紅色的分佈多達40%外,其餘部 份之分佈和整體分析中之分佈相似。黃苔分佈,舌中百分比分佈有到50% 是最多;舌的兩邊到30%是其次;在舌尖的比例分佈在0∼10%之間,幾乎 沒有黃苔的分佈。白苔分佈情形與黃苔相似,但較大。在系統限制方面, 運算速度及分割技術仍有待進一步改進。 由以上結果顯示,本計畫之系統已具備初步舌診的功能,未來希望能 配合中醫師蒐集更多臨床資料,擴大樣本之多元性,對系統再行修正,並 建立資料庫,如此才能達到初步的目標--運用電腦來做影像處理分析,提 供客觀的數據,做為中醫診斷時之參考,並成為中醫舌診科學化之基礎。 |
英文摘要 | Tongue Diagnosis is a important item in observational determination of Chinese medicine diagnosis. Chinese medicine doctors mainly observe the color, movement, shape and properties of tongue when they are diagnosing. It's a tread that the quantitative analysis via computerlized Chinese tongue diagnosis in recently years. The purpose of this study is to recognize the color and shape of tongue, color of coat, by using color image processing techniques. A system was established for the quantitative analysis of tongue diagnosis by hardware and software integration. In the study, the tongue image was grabbed and calibrated, the tongue characters were distinguished via our system. In pre-processing of tongue image, the calibration of environmental factors was emphasized, including global calibration and RGB standard calibration. The results that it could compensate image deviation which caused by different environments. In the analysis of tongue shape, the edge of tongue image was obtained then the length, width and area of tongue could be calculated by using boundary detection method. Also, the thickness of tongue was recognized by using stereoscopic imaging techniques. For the development of algorithms to analysis the tongue's characters, tongue color and coat color were recognized by using segmented tongue image which were take from 15 health people. Besides, analyzing and recognizing of whole tongue image, the tongue image was partitioned to four parts to analysis the distributions of tongue characters according to the principles of Chinese tongue diagnosis. After the system was constructed, the property of the distributions of tongue phenomenon and the regularity of tongue coat variation were generalized by testing 50 health men cases. In the global analysis of Tongue characteristic distribution, the percent of the color of tongue was more then 50% which light red tongue is beyond 410% and red and dark red tongue is less then 10%. In the distriubtion of the color of coat, white coat was almost beyond 20% and more then yellow coat. The result show that less than 0.5% case was without decision based on threshold histogram method for distinguishing yellow coat and white coat. Most of tongues were 4-5 cm for the width, 4-6 cm for the length, and 15-25 cm�惠or the area of the tongue by using the segmentation of image method. The distribution of tongue characteristics in sub-regions, the color of tongue and the color of coat which located in the middle of tongue were not different, but distribution on the left and right region were 40% and the top of tongue are located almost within 20%, which was smallest. In sub-regions of the distribution, red and dark red tongue are almost within 10%, but top of red tongue increased to 40%, but the distribution of rest part was similar with distribution in global distribution. The distribution of yellow coat, 50% is the largest in the middle tongue, 30% are largest in the left and right tongue and 0 - 10% for the top of tongue. The distribution of white coat are likely to yellow coat, but which are more then yellow coat. In system restrictions, the speed of calculation and technique of segmentation got to be improved in further. The result showed that our system supported achievement of a primer of tongue diagnosis. In order to accomplish primer purpose, more variation sampling should be obtained, this system was modified, then system database was established. Besides, analysis and process the image by using computers would provide the objective numerical data to support the science evidence for Chinese tongue diagnosis. |
本系統中英文摘要資訊取自各篇刊載內容。