A hybrid defect detection method for compact camera lens

Abstract

Production technology has increased rapidly with the development of industrial technology. Conventional human visual inspection is insufficient for conducting quality control under this increased capacity. Therefore, high-speed and high-accuracy automated optical inspection is becoming increasingly crucial. In this article, we propose an automated inspection method for a compact camera lens using a circle Hough transformation, weighted Sobel filter, and polar transformation. Our analysis of defects in the compact camera lens identified problems including of the circular texture and the non-fixed position of the inspection region. To overcome these problems, we design an inspection algorithm for locating and inspecting a circular region. A machine learning support vector machine method is then applied for obtaining a precise detection result. The experimental results show that the proposed inspection method is suitable for detecting defects in a complicated circular inspection region, and that the proposed system exhibited high performance.

A Review and Analysis of Automatic Optical Inspection and Quality Monitoring Methods in Electronics Industry

“…Minor defects were also included in separate category. [290] IR-CUT filter defects such as stain, scratch, and edge crack [296] Surface defects in micro multi-layer nonspherical lens module of CMOS such as bright spot, dark spot, scratch, foreign material and hole [297] Compact camera lens and spacer ring defects such as stain, bright dot, scratch, pit, and scar Passive Components [291], [292] Ripple defects in the surface barrier Layer chips of ceramic capacitors [298] Tiny surface defects in the surface barrier Layer chips of passive electronic components [299] Surface defects of film capacitors Thermal Fuse [293] Bur, black dot, small-head, and flake defects and lightning setup, computer (processor), conveyor and sorting mechanism as shown in Figure 16. The illumination is responsible for providing constant/customized lightning conditions.…”

“…Despite of the proposed method's ability to outperform the conventional one; however, it is only capable of detecting nonstationary straight-line and cannot consider nonstationary curved profiles and nonstationary linear/curved surfaces. Chang et al in [297] used Hough Transform as a preprocessing approach to transform the circular shape of camera lens' into a linear shape before inspecting the defects. After the transformation from the polar coordinate into Cartesian coordinates, segmentation and morphological operations were conducted to segment the defects.…”

“…k-NN algorithm was not widely used in the research articles reviewed in this paper. However, some articles such as in [12], [73], [91], [116], [143], [153], [154], [171], [297] used k-NN to compare its performance with the main classi-This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication.…”

Intelligent manufacturing/production systems: Modeling, algorithms, and optimization

“…In the second category, intelligent manufacturing systems, Chang et al 8 design an algorithm using machine learning to support fault detection in camera lens production. The proposed approach demonstrates the strength of applying machine learning to standardized processes, such as quality control.…”

Apparatus and Method of Defect Detection for Resin Films

“…After extracting the features, an SVM was proposed to classify defects. Chang et al [17] proposed a method for defect detection on the compact lens. He segmented objects by applying weighted Sobel filters and watersheds and then used the SVM for classification.…”