Research On Feature Extraction And Classification Recognition Of Fracture Images

The automatic interpretation system of X-ray bone images based on intelligent identification uses computer to process and analyze fracture images. It proposes the appropriate understanding strategies of fracture images, achieves the automatic interpretation of fracture types and injury degree. The study and application of automatic interpretation can help doctor relieve from heavy works of reading X-ray fracture images, give support to doctor on diagnosis and treatment of fractures and improve the efficiency and accuracy of diagnosis. The technique of feature selection,feature extraction and fracture classification is the key technique of automatic interpretation of fracture images. Feature selection and extraction is the main step to achieve classification of fracture images. Classification of fracture images is the main purpose.This paper makes X-ray fracture images of femoral shaft as study objects, proposes the morphology features extraction algorithm of femoral shaft fracture images. First, transforms the visualized segmentation results of femoral shaft fracture images into binary images. Later, makes boundary tracking on binary images to determine the edge points of fracture images; Finally, extracts morphology features: region number,region area,region centroid,protuberant polygon of fracture images,centerline and fracture line of every region based on region centroid,computes the value of rectangle degree,circularity degree and the angle between fracture line and perpendicular line of centerline.This paper makes the table of decision system based on the extracted feature parameters, applies C4.5 algorithm to the classification of X ray images of femoral shaft fracture. It inducts decision support degree as the test-property according to the instability of information gain ratio as the test-property. The decision support degree of a property is the supporting degree to get the correct classification to use this property to classify samples. It gives the recognition algorithm of classification based on decision support degree, achieves the classification of femoral shaft fracture images and attains the classification codes. Finally, determines the types of femoral shaft fracture and selects treatment programs of fracture images according to the classification codes.