人體影像異常區域統計檢測方法

The invention relates to a statistic measure for abnormal area medical analog image of human body and, more especially, to the image composed of each signal point intensity such as infrared thermal image. The present invention can be used in human body medical image measurement software. A group of whole or partial human body image standard database which has been processed by image lofting and handling procedures is used for focusing on the measure range of gradually gathering statistics of image grey distribution curve of database therein. In addition to producing image-based statistic result, such as visual data related image grey level statistics including average, standard deviation, middle value, mode, skewness, kurtosis, etc., manual input of overall normal probability percentage is employed to perform upper-bound and lower-bound calculations of qualified image grey level on a normal distribution curve of the image grey level point by point, so as to classify all image pixels of the patient's affected part into normal type and abnormal type based on the upper and lower levels. Such abnormal pixels usually present dispersed point distribution on geometric distribution. After the process of image pinhole erasing and decontaminating, dispersed image is rearranged into a whole abnormal image area suitable for a doctor's observation, and finally the goal of medical diagnosis of judging geometric location of sick organ is achieved. The best advantage thereof is to provide image-based statistic data and to remove the difference and preserve the similarity in the normal image database. Further, a computer compares and rearranges automatically the abnormal pixels of grey image and quickly recognizes abnormal image area finally.