Processing And Detection Of Industrial And Medical X-ray Image

Now, X-ray is widely used in industrial non-destructive testing and medical perspective inspection because of its strong penetrability. Unfortunately, the X-ray non-destructive testing systems in both industrial and medical fields are still done by human interpreter and these methods are easily affected by devices, environment and people's physiological conditions. Therefore, with the development of image processing and detection technique, a computer-aided method needs to be found to make detection normative and intelligent. As the reasons above, some new methods that can process and detect X-ray images automatically are studied in this paper:1. Taking the X-ray detection image as the research object, this paper introduces an algorithm based on SVM (Support Vector Machine) that can detect the defects in the industrial products automatically. Firstly, the images are preprocessed to enhance their qualities. Secondly, a number of sample images are collected and the contrasts of these sample images are extracted as feathers, which are input to train the SVM. Finally, the features of the detected images are extracted and put into the trained SVM to detect the defects in the images.2. When a 12-bit CT image is displayed on a common 8-bit computer monitor, a lot of gray details will be lost. Aiming at the problem above, a new automatic display method for CT images based on region detail extraction and image fusion is presented. Firstly an original 12-bit CT image is segmented into several regions by FCM (Fuzzy C-means) algorithm and each region represents a kind of correlated human tissue. Then in order to extract the details in these regions, every single region is treated by a corresponding display window. As a result, several 8-bit images are extracted from the original 12-bit image. Finally, all these detail images are fused by using the local variance weighted method, and an 8-bit result image which contains more details of human tissues is obtained.3. A digital film management system is realized, where the detection algorithm based on SVM is embedded.In order to confirm the validity of the methods in this paper, several defect images obtained from the factory and several 12-bit human brain CT images are utilized to test the algorithms in this paper. The results show that the methods in this paper are valid and satisfied.