Search results

The detection of invisible micro cracks (μ‐cracks) in multi‐crystalline silicon (mc‐si) solar wafers is difficult because of the wafers' heterogeneously textured backgrounds. The difficulty is twofold. First, invisible μ‐cracks must be visualized to imaging devices. Second, an image processing sequence capable of extracting μ‐cracks from the captured images must be developed. The purpose of this paper is to reveal invisible μ‐cracks that lie beneath the surface of mc‐si solar wafers.

To solve the problems, the authors first set up a near infrared (NIR) imaging system to capture images of interior μ‐cracks. After being able to see the invisible μ‐cracks, a region‐growing flaw detection algorithm was then developed to extract μ‐cracks from the captured images.

The experimental results showed that the proposed μ‐cracks inspection system is effective in detecting μ‐cracks. In addition, the system can also be used for the inspection of silicon solar wafers for stain, pinhole, inclusion and macro cracks. The overall accuracy of the defect detection system is 99.85 percent.

At present, the developed prototype system can detect μ‐crack down to 13.4 μm. The inspection resolution is high but the speed is low. However, the limitation on inspection speed can easily be lifted by choosing a higher resolution NIR camera.

Generally, this paper is a great reference for researchers who are interested in developing automatic optical inspection systems for inspecting solar wafer for invisible μ‐cracks.

The research described in this paper makes a step toward developing an effective while low‐cost approach for revealing invisible μ‐crack of mc‐si solar wafers. The advantages provided by the proposed system include excellent crack detection sensitivity, capability of detecting hidden subsurface μ‐cracks, and low cost.

Though many segmentation methods have been published, few of them are developed especially for line scanned images. An ill‐illuminated line scanned (IILS) image tends to have a uniform intensity distribution in column direction while non‐uniform intensity distribution in the row direction. So, it is improper to segment IILS images using either a pixed threshold or threshold surface. In view of this, the purpose of this paper is to develop a segmentation method that is suitable for segmented IILS images.

To obtain satisfactory segmentation results, the illumination variation across the column of a line scanned image was taken into account and a column‐based segmentation method was developed. The method first calculates each column's standard deviation. Then a threshold value is automatically assigned to each column based on the derived values. Finally, by assembling each columns threshold value, a so‐called threshold line is formed. The method is threshold‐line segmentation method based on standard deviation (TLSTD).

The developed threshold‐line‐based segmentation method is compared with Otsu's fixed threshold segmentation method and Niblack's threshold‐surface‐based segmentation method. The results show that the threshold‐line‐based segmentation method is more suitable for segmenting IILS images.

Despite TLSTD outperforming Otsu's and Nilblack's segmentation methods, there are some limitations to it. The most obvious one is that the predetermined allowable deviation has influences on the integrality of the extracted flaws. Besides, since the proposed method is designed specifically for segmenting images captured by line scan cameras with a slant line light source, it is suitable for segmenting the kind of images only. In other words, the method shows no advantages in segment area scanned images.

Generally, the approach is useful in automated visual inspection where line scan cameras are employed.

The merit of the proposed method is that the slant of the line light source is now allowed. In other words, even if a grabbed line scanned image is unevenly illuminated, the proposed segmentation method is still able to successfully detect desired flaws.

The purpose of this paper is to develop an effective and reliable corner detection algorithm so as to extract all the desired corners successfully. In addition, the influences of edge detection method as well as smoothing technique on the overall performance of corner detection techniques are investigated.

In this paper, an effective corner detection algorithm based on subpixel edge detector and Gaussian filter is presented. First, a subpixel accuracy edge detector is used rather than a pixel accuracy edge detector to detect edges. Second, B‐splines approximation technique is used to eliminate the staircase effect of a digital curve. Third, curvature curve derived from the edges is smoothed by a Gaussian filter. Finally, statistical process control technique is applied to detect vertices.

The results show that spatial‐moment outperforms chain code as an edge detector. Furthermore, the Gaussian filter should be used to smooth curvature curve instead of smoothing the profile of an object, because the former provides greater impact on the corner detection results.

In addition to object recognition, motion tracking and obstacle avoidance, the proposed method also has many important engineering and manufacturing applications such as dimensional measuring, reverse engineering, and machine vision‐based computer numerical control (CNC) machining of polygonal sheet metal parts.

The purpose of this paper is to present an automatic inspection method of colors and textures classification of paper and cloth objects.

In this system, the color image is transformed from RGB model to other suitable color model with one of the components being chosen as the gray‐level image for extracting textures. The gray‐level image is decomposed into four child images using wavelet transformation. Two child images capable of detecting variations along columns and rows are used to generate 0° and 90° co‐occurrence matrices, respectively. Some of the distinguishable texture features are derived from the two co‐occurrence matrixes. Finally, the test image is classified using neural networks. Nine color papers and eight color cloths are used to test the developed classification method.

The results show that recognition rate higher than 97.86 percent can be achieved if color and texture features are both used as the inputs to the networks.

The paper presents a new approach for testing materials. The multipurpose measurement application with unsophisticated and economical equipment can be confirmed in online inspection of papers and cloth manufacturing.