首页> 美国政府科技报告 >Some Considerations on Quality Aspects of Low Level Image Processing
【24h】

Some Considerations on Quality Aspects of Low Level Image Processing

机译:关于低级图像处理质量方面的几点思考

获取原文

摘要

This thesis deals with quality aspects of low level image processing. In order tocomputerize processes adequately, it is necessary to have insight into these aspects. Specifically, there is studied on the relation between the quality of input data and the quality of the results following from a basic gradient operator. An important characterization of the quality of input data in digital image processing is the amount of noise which is present in an image. Therefore, a noise estimation method proposed by Forstner is investigated. The noise estimations and the propagation of the noise on the gradient functions are used for the statistical determination of thresholds. Thresholding can be used for the detection of edges in an image. Pixels with a gradient function value higher than a considered to be non-edge pixels. Obviously, thresholding will result in classification errors. The misdetection of an edge pixel is a first order error, and the detection of a non-edge pixel as edge pixel is a second order error. If the statistical distribution of edge and non-edge pixels is known, a threshold with optimal errors can be determined. The analysis of isotropically distributed mixed-pixel edges is also applied to a real image. The amounts of contrast are guessed from the gradient histograms of the image and the thresholding output is compared with the output of a straight-edge detector. The results are a good base for the further investigation of thresholding errors in real images.

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号