Quantitative defects detection has always been the one of the difficulties in optical element surface quality evaluation. In order to solve this problem, the optical element surface defects detection based on dark-field imaging system, which has been researched by our group team for nearly twenty years, has been summarized. The plane and sphere optical element surface defects detection details are introduced. Specifically, it involves plane optical element surface leveling, sphere optical element spherical center alignment, low magnification image acquisition, low magnification image stitching, feature extraction, high magnification defects detection and report output based on the form of specific standard (Such as America Military Standard MIL-PRF-13830B or China National Standard GB/T 1185-2006). Besides, a China National Standard about digitized quantitative measurement of the defect, which is proposed by our group (now is in the stage of request for public advice), is also introduced.
Surface defects inspection is a critical part in manufacturing of mobile phone cover glass. Considering the defects in the optical elements surface caused by the imperfection of manufacturing technique, the classification and the position information should be carried out for the necessary repairing process. The traditional manual inspection method is always labor-consuming and inefficient. Surface defects digital evaluation system based on machine vision draws much attention in the recent years. This paper proposed algorithms and applications for the detection task with higher efficiency and reliability comparing to the manual inspection. The detection method is based on machine vision and machine learning techniques. The images of optical elements surface are captured by line-scanning cameras, with the imaging systems of dark-field, bright-field and transmission-field. Only one image system is not enough to detect all kind of defects like scratch, bubble, crack of glass and edge chipping etc. The position information and category of defects are obtained based on image processing technique. The defective area was calculated by image filtering algorithm, the feature selection techniques based on segmentation methods are explored and the feature vector can be extracted before the next step of classification with Support Vector Machine (SVM) technique. Verified by the experiments, the results reveal this method has good performance and is very suitable for recognition and classification of glass defects.
Aimed at the problem of strong background interference introduced in digital image processing from complex surfaces under industrial defect detection, a method for complex surface defect detection based on human visual characteristics and feature extracting is proposed. Inspired by the visual attention mechanism, defect areas can be identified from the background noise conveniently by human eyes. We introduce the improved grayscale adjustment and frequency-tuned saliency algorithm combined with the salient region mask obtained by dilation and differential operation to eliminate the background noise and extract defect areas. Meanwhile the directional feature matching and merging algorithm is applied to enhance directional features and retain details of defects. Testing images are captured by our established detecting system. Experimental results show that our method can retain defect information completely and achieve considerable extracting efficiency and detecting accuracy.
In inertial confinement fusion system, the intermittent scratches on the polished surface of single-sided polished and bottom surface frosted optical components are complex, and it’s of great difficulty to extract them completely. In order to solve this problem, established in the light-field surface detection system, this paper brings forward a novel intermittent scratch detection method based on adaptive sector scanning algorithm (ASC) cascading mean variance threshold algorithm (MVTH). In the preprocessing step, dividing the original image into subimages with a number of integer multiple of cpu cores so as to fully compress image processing time utilizing parallel processing, using mean filter to balance background and then obtaining binary subimages utilizing morphology and threshold operations, finally, utilizing Two-pass algorithm to label the connected domains of binary subimages. In the detection step, considering the complexity of the pattern of intermittent scratches, ASC is first used for routine intermittent scratches stitching and then supplemented by MVTH. In the verification step, in order to prove that the detected intermittent scratches satisfy the criteria for scratches in human eyes, the method of support vector machine (SVM) pattern recognition is utilized to compare the detected results with the continuous scratch samples detected by human eyes. This algorithm has high degree of parallelism, high speed and strong robustness. The experimental results illustrate that the complete extraction rate of intermittent scratches is 93.59% , the average processing time of single image is merely 0.029 second and the accuracy rate of detection is up to 98.72% by SVM verification.
In view of the difficulty of defects detection of complex metal curve surface in uneven illumination and high speed processing, a new, simple, yet robust algorithm based on statistical feature of local visual field is proposed. This algorithm first performs the ideal image difference by extracting the template from the image itself, and then computes the statistical feature in local visual field to correct the gray-scale fluctuation in each region of image. In this way, the influence of the uneven illumination at low and high frequency is eliminated concurrently, which achieves the equalization of the statistical features of the local visual fields except the position containing the defect, so as to use the global threshold in whole image reasonably; Next, on the search of defects, this paper replaces the pixel level with the local field of vision and compresses the image information with the defects’ scale which is in line with the human eye. This not only reduces the influence of random noise, but also greatly improves the processing speed while preserving defects information, which makes it possible to realize real-time processing ability for image with the large amount of data. To detect complex curved surface on semi-finished metal shell of cell phone, the experimental results demonstrate that the defects detection accuracy of the proposed algorithm can reach 95%, and the detection time for single test area is less than 1ms, which is suitable for accurate and real-time detection on the production line for such surface defect.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.