Computational tools for medical image analysis help clinicians diagnose, treat, monitor changes, plan, and execute procedures more safely and effectively. Two fundamental problems in analyzing medical imagery are image registration and image segmentation. Analysis of Cephalometric x-ray and measurements of parameters based on detected features play an important in monitoring the treatment. The analysis is based on location of landmarks that form the basis of linear and angular measurements. Many systems had developed to computerize cephalometric process but most of them didn't satisfy user requirements in accurate, speed, and reliability. In this dissertation a new system is developed to automate cephalometric analysis and measurements using x-ray images. For cephalometric systems our study consists of three main parts: filtering model, locating landmark model, and measurements model. The use of filter on x-ray can increase cephalometric system accuracy. Radiographic images contain many features. Some of them are easily distinguishable while others are more robust when filter is considered. A particular method of defining Gaussian and unsharb filters appropriate for x-ray images is used. The filters are adapted to the domain of radiographic images. Two alternative approaches are implemented to provide suitable filters for enhancing the x-ray images. These approaches use Gaussian filter and unsharb filter to provide powerful tools for locating the important landmark points in x-ray image to obtain better accuracy in cephalometric measurements. A new system is developed for cephalometric analysis, measurements and locating of landmark points. Constructions of landmark points are done by manual placement of a number of nodes with respect to a registered x-ray image. Locations are chosen by experienced dentist who selected the areas of distinctive appearance on images of different skull. Eleven nodes are defined on the developed system. The nodes are very carefully placed to ensure that their placement was at the intended point on the x-ray image. At each node location, a set of (x,y) point of the coordinates are registered for measurement purpose. Analysis and measurements processes are performed automatically, including 18 measurements of distance and degree for x-ray image. Statistical analysis shows better accuracy for the developed cephalometric system than manual methods. The testing results demonstrated that filter methods can indeed provided a significant increase in accuracy for locating landmark points. This new system is fast and superior in accuracy compared to the manual cephalometric methods. There is no static difference between manual and automatic measurements. The experimental results have demonstrated that the proposed method performs better than the existing methods in terms of their identifiably, accuracy, speed, and dynamic range of images. A prototype was developed with a Multi-Document Interface (MDI) application of Microsoft Windows. Three modules were included. â€¢ Manual landmarking module provides the function manual to locate the landmarks on the screen through the interactive user interface. â€¢ Filter module enhances the contrast of x-ray images. â€¢ Measurements are done in the cephalometric analysis module. The system has proved to be useful tool for orthodontists in the diagnosis and treatment of cephalometric analysis.

An image processing approach for cephalometric measurements

/pdf/analyse-d-images-radioscopiques-et-fusion-d-informations-a6ic3j1nfv.pdf

Analyse d'images radioscopiques et fusion d'informations multimodales pour l'amelioration du controle de

This paper aims at the problems of identification of cheese yarn in low resolution pictures. On this basis, a cheese yarn label identification system based on machine vision technology is proposed. By using the improved Hough algorithm, the background of the label is removed. On the basis of transforming images from RGB color space to HSV color space, the color-histograms of the label are obtained. Then, the weights of the neural network are obtained by training the neural network with histogram information, and the neural network is used to judge whether the types of yarns are the same as the samples. As verified by experiments, the stability and accuracy of the method can meet the practical application requirements and could effectively identify the types of cheese yarns.

Study on the Type Identification of Cheese Yarn Based on Low-Resolution Pictures

A comparison is made of the conventional difference-of-Gaussians (DOG) filter and a morphological filter based on opening with a spherical structuring element in realistic images from X-ray inspection. The task is the detection of bright blobs in low-signal/high-noise images containing other objects, such as step edges and corners. The study shows that the morphological filter has a superior selectivity, as it is insensitive to edges, but the resulting images are not free from extra signals. Postprocessing is required to remove undesired responses at sharp, bright corners and similar narrow structures. The morphological filter is more susceptible to noise than its linear competitor, since no averaging is performed. Presmoothing can alleviate this problem, but the choice of the smoothing filter is nontrivial if the advantages of the morphological filter are to be preserved. >

H. Boerner

Papers

Feature extraction by grayscale morphological operations-a comparison to DOG filters