Der-Baau Perng

This paper describes a global image restoration scheme using a discrete cosine transform (DCT) that can be used to detect defects in directional textured surfaces automatically. The input spatial domain image is first transformed into the DCT domain so that the dominant directions of the texture in the input image will be compacted into the orthogonal straight lines or impulses through the direct current (DC) component of the spectrum. The linear primitives associated with the high-energies in the DCT domain are eliminated by reducing them to zero before transforming back to the spatial domain. Finally, the defects, if any, are extracted by the thresholding method. Experiments on a variety of product surfaces with directional textures such as straight, slanted, orthogonal, slanted orthogonal, and oblique linear primitives were conducted in this paper. The proposed scheme would blur directional textures and preserved only local defects if they were initially embedded in the image. Furthermore, some preliminary experiments were also conducted to demonstrate the proposed scheme was insensitive to horizontal and vertical shifting, changes in illumination, and image rotation.

This article describes a novel inspection system that uses machine vision to detect optical defects in quasi-contact lenses. The hardware consists of a host computer, a die float, and an interferometer that produces grating moiré fringe pattern images of the quasi-contact lenses being inspected. The major optical region of the lens image is first segmented, and then the middle axes of each fringe on the optical region are determined. Three characteristic features of the fringe, namely, the degree of twist, parallelism, and orthogonality, are extracted to create an effective mapping of the original features to a semantic description of the textures. Finally, the quality of the quasi-contact lens is determined using a control chart procedure. A set of 147 quasi-contact lens images was used to validate the proposed system. The successful detection rate was 95.3%, demonstrating that the proposed algorithm is a feasible solution for detecting optical defects on the quasi-contact lenses.

No available method can automatically verify the correctness of the wire bonding positions on a multi-layered wire IC. This paper presents a novel method that integrates image processing and wire bonding simulation techniques. The proposed method first takes the IC leadframe image and calculates the lead information before actual wire bonding begins. The wire bonding position information is then generated to simulate the actual wire bonding process. The generated pseudo bonding information is then compared with that from a referential machine. This approach can check the wire bonding position correctness before any actual wire bonding is executed. This approach can fully solve the mal-detection and lost detection problems that may occur in other available methods. The experimental results show that the proposed bonding position check (BPC) method is robust and fast enough for applied multi-layered wire IC inspection synchronously with the wire bonding process.

In this paper, an innovative automated optical inspection (AOI) system is proposed to detect fishtail collapse of various fishtail images. Up to the present, there is no report on the inspection of fishtail collapse by using an AOI system. The proposed AOI system is the first system that can inspect the fishtail collapse of microrouter. The fishtail image of the microrouter was extracted first by a newly developed lighting device. We use the end gash to identify one of the blades as the reference blade, in order to provide an orientation-invariant method to solve the fishtail image rotation problem. Three invariant features of the blade are used as indicators of quality control. The blade with collapse, whose area is as small as 0.02 mm × 0.02 mm, could be identified based on the constructed quality control charts. Twenty-five fishtail images are used to validate the proposed AOI system. The successful detection rate of the implemented AOI system is up to 99.2%, and it demonstrates that this system detects the collapse accurately and robustly.

A novel lighting system was devised for 3D defect inspection in the wire bonding process. Gold wires of 20 microm in diameter were bonded to connect the integrated circuit (IC) chip with the substrate. Bonding wire defects can be classified as 2D type and 3D type. The 2D-type defects include missed, shifted, or shorted wires. These defects can be inspected from a 2D top-view image of the wire. The 3D-type bonding wire defects are sagging wires, and are difficult to inspect from a 2D top-view image. A structured lighting system was designed and developed to facilitate all 2D-type and 3D-type defect inspection. The devised lighting system can be programmed to turn the structured LEDs on or off independently. Experiments show that the devised illumination system is effective for wire bonding inspection and will be valuable for further applications.

We devised two sets of vision systems which can automatically inspect defects in CRT panels. One system inspects the front surface defects while the other system inspects the CRT panels' side-wall defects. The defects of a 3-D translucent hull CRT panel may include scratches, blemishes, cracks, bubbles, or collapsed portions which may occur on the inside surface, outside surface, or interior of the glass. Two different kinds of lighting environments were developed to highlight the CRT panels and to simplify the image-capturing process. Meanwhile, two sets of methods were devised which can reliably find all the defects within the expected processing time. Experimental results demonstrate that the developed vision systems can be used to effectively and efficiently auto-inspect CRT panels in the production line.