Dissertation > Excellent graduate degree dissertation topics show

Correction Algorithm for Fiber Images with Non-Uniform Illumination

Author: DongXinYou
Tutor: ZengPeiFeng
School: Donghua University
Course: Computer Applications
Keywords: Iterated function systems Mean Filtering Window function Fiber identification Image binarization
CLC: TP391.41
Type: Master's thesis
Year: 2008
Downloads: 32
Quote: 0
Read: Download Dissertation

Abstract


Fiber image analysis and automatic identification system, we need a close-imaging technology. However, in the actual image acquisition process, due to the limit of the microscope, the imaging system, the image acquisition system, a variety of environments, such as, the use of a simple point source, such a light source system provides a non-uniform lighting in the central portion of the image is the direct angle of deviation in the other part of the light occurs, so that the image appears on the uneven illumination. The non-uniform illumination in the background noise generated in the image, and image signals are mixed together, causing the contrast of the image and the gray scale distribution uneven. This distortion affect the performance of the fiber recognition processing algorithms. Therefore, the required correction algorithm to eliminate the uneven illumination introduce the background noise. Effectively remove uneven illumination generated background noise at the same time, useful information on the image retention is the evaluation the illumination unevenness correction is an important feature of the performance of the algorithm. Uneven illumination caused by background noise and the image of useful information there is overlap in the image domain and frequency domain, and therefore must be discussed to eliminate the uneven illumination useful component of loss. This article first image uneven illumination algorithms. Depending on the uneven illumination correction method is divided into two types of image domain and frequency domain method that is based on the image domain uneven illumination correction technology based on frequency domain filtering uneven illumination correction technology. Transform method, the former on the image where the space for processing, the image domain refers to the space, image composed of pixels temporal enhancement method is a direct role in the enhancement method of the pixel; the latter to the processing of the image by the Fourier transform, wavelet transform, etc. The image is converted to the frequency domain by a different analysis of the frequency of the image signal processing operations. However, a variety of different treatment has its own inadequacies. The effect of the image domain in order to get better uneven illumination correction, large amount of calculation; frequency domain filtering in image edge blurring effect and filter side effects. And uneven illumination correction algorithm is still a lack of a unified theory of evaluation, which is not corrected mass general objective standard measure uneven illumination. Correction methods often targeted the correction results after just rely on the subjective feeling of the people to be evaluated. In order to overcome the shortcomings of the above two methods, the illumination correction algorithm based on linear iteration. First, the image signal deviation from average linear iteration. In this iterative process, by calculating from the average difference deviation from the local average amplitude structure set of feature points, and the corresponding intensity value of the local intensity of the feature point position average substituted, constantly eliminating severe change in the edge of the intensity value of the intensity value changes slowly the edges of the impact. Ultimately achieve the effect of uneven illumination correction. Experimental results show that the linear iterative algorithm can remove uneven illumination of the background noise on the image, can be a good solution to the problem of imbalance of light noise in the image edge detection and local strength values ??compared, and the calculation speed.

Related Dissertations

  1. Research on Identification System of Cashmere and Wool Fiber,TS101.921
  2. A Study on Scattering from Target and Rough Surface by Using FDTD Method,TN011
  3. Research on the Entropy-based Music Fingerprinting System,TP391.3
  4. NVD based on fractal pattern background animation audio playback,TP391.41
  5. Phase-coded signal pulse technology research and implementation,TN957.51
  6. Sub -band Blind Separation of Speech,TN912.3
  7. Study on Filtering Algorithm of the Aerial Gravity Measurement Data Processing,P223
  8. Fractal Theory and Its Application in Financial Market Analysis,F830.9
  9. Research on Processing of Range Information of Shore-Based Isar,TN957.51
  10. Video text extraction technology research and application,TP391.41
  11. Based on Object-oriented Programming Graph Arrow Quantification Research,TP391.41
  12. The Study of QR Code Decoding Algorithm on the Android Mobile Platform,TP391.41
  13. The Codingalgorithm for Image Encryption and Compression Based on Chaos and Fractal,TP391.41
  14. Study on the Key Technique of All Optical Two-dimensional Negative Feedback System,TP391.41
  15. Study on Eddy Seam Tracking Sensor Characteristics and the Effecting,TP212
  16. The Design of Measuring System for Mining Methane Catalytic Sensor,TP212.2
  17. Research on Image Denoising with Wavelet Transform,TP391.41
  18. The Design of Tank Truck Loading Devices Positioning System,TP391.41
  19. Window function based on a combination of HHT in the end effect of inhibiting the exploration,TN911.7
  20. The Iterated Function System of a Class of Multi-parameter Recursion Fractal Interpolation Surface and Its Box-dimension,O186.1

CLC: > Industrial Technology > Automation technology,computer technology > Computing technology,computer technology > Computer applications > Information processing (information processing) > Pattern Recognition and devices > Image recognition device
© 2012 www.DissertationTopic.Net  Mobile