On Methods Of Digital Image Segmentation And Quality Evaluation

The human society marched into the information age. As the foundations of the information society, science computation, communication and network technology, have obtained the swift and violent development, popularizes day by day along with these technical unceasing development and are enormously influencing people's life style and even the entire society's production method. The carrier to exchange and transmit the information mainly is the visual information, namely image information. The image information demand has led to image correlated technology development, the image segmentation and the image quality evaluation are one of these important technologies.This thesis has conducted the system research to the current tradition segmentation method, and proposed a new method using the morphology method and the ellipse nature to extract the ellipses in images The robust ellipse extraction has specially vital significance in the pattern recognition and the computer vision, because including the circle is the simplest closed curve, it is the most common curve element in geometry pattern. To conduct the image pretreatment using the advantage of morphology in edge extraction, then to extract ellipse using the ellipse culmination and extreme - string nature. Has confirmed this method through the simulation experiment more simple compared with other previous methods. It is more effective to extract only ellipses in images.The output image is for people's observation and evaluation, thus the research of image. quality evaluation is also especially important. This thesis puts forward a quality assessment method based on the morphology gradient by the systematical research to the existent image quality assessment methods.The morphology gradient used in this method is different from traditional gradient operator which is easy to strengthen the noise, moreover in view of the low noise sensitivity characteristic, filters the noise influence in the gradient image, at the same time filters small gradient value which caused image blurring by selecting threshold value. Through acquiring the gradient of the processed image and image quality evaluating, it is proved that this appraisal result tallies with the subjective appraisal much better, compared with other gradient operators and methods.