Font Size: a A A

The Key Processing Technologies Of Microscopic Video Stream And Its Application

Posted on:2016-04-29Degree:DoctorType:Dissertation
Country:ChinaCandidate:Y LiuFull Text:PDF
GTID:1108330464954333Subject:Optical Engineering
Abstract/Summary:Request the full-text of this thesis
With the rapid development of information science and electronic technology, traditional optical microscope already cannot meet the demands of people. Modern digital microscope use electronic eyepiece to replace the eyepiece of traditional optical microscope, the electronic eyepiece of which can be connected with computer or other display devices. The digital images can be captured to do further post-processing and data analysis by the powerful microscopic image processing software. At present, there are significant differences between the microscopic image processing software produced in domestic company and the international giant microscopic enterprises. Our research realized some key advanced microscopic image processing function based on image processing and computer vision knowledge. In addition, the microscopic system were applied to the industry area, and we provided the solution for the automatic optical IR-CUT defects inspection system.In the respect of microscopic video stream output, this paper designed the pipeline of color image processing for the electronic eyepiece, including black level adjustment, defective pixels removing, linearization, white balance, color filter array (CFA) interpolation, color correction, tone adjust and gamma correction. In this paper, we narrated the basic microscopic image enhance operation and compared different de-noising schemes based on stationary wavelet transform. In addition, we proposed the signal to noise ratio (SNR) testing method of cooled electronic eyepiece based on the stationary wavelet transform.In the advanced microscopic image processing algorithm part, we designed the automatic and manual microscopic cell/particle count algorithm, and proposed the image fusion algorithm for unregistered multi-focus/multi-exposure image sequence. We highlighted the registration and fusion process of unregistered multi-focus image sequence. Speeded Up Robust Features (SURF) feature detector with Binary Robust Invariant Scalable Keypoints (BRISK) feature descriptor matching scheme is used in the registration process of multi-focus image sequence. An improved RANdom Sample Consensus (RANSAC) algorithm is adopted to reject incorrect matches. The registered images are fused using stationary wavelet transform with sym5 wavelet basis. The experimental results prove that the proposed algorithm achieves better performance for unregistered multiply multi-focus images, and it is especially robust to scale and rotation translation compared with traditional direct fusion method.In the application of microscope system, we presents an automatic surface defects inspection system for optical IR-CUT filter, which involves illumination and imaging module, moving module, flipping module and machine vision algorithm. There are three kinds of defects may exist in the surface of optical IR-CUT filter:stain, scratch, edge crack. To highlight all the defected regions on the optical IR-CUT surface, an improved dark-field illumination technique is utilized in the imaging module and two infrared ring form LED (light-emitting diode) light sources with different incident angles are chosen as illuminant for the characters of optical IR-CUT filter and the defects. Meanwhile, monochrome CCD (Charge Coupled Device) camera is embedded in the system as an imaging module. In order to localize the region of optical IR-CUT filter in the captured image accurately, stationary wavelet transform is introduced to template matching algorithm. A new threshold method using SWT and sobel operator is applied to the extraction of defects, which avoids the use of complicated learning process from a set of samples. Experimental results on a variety of optical IR-CUT filter samples, including good samples, samples with defects of stain, scratch and edge crack, have shown the effectiveness and accuracy of the proposed system. Convexity theory is implemented on the algorithm of defects classification for edge crack.
Keywords/Search Tags:microscopic image processing, color image processing pipeline, cell count, multi-focus image fusion, defects inspection
Request the full-text of this thesis
Related items