Lens Inspection System Using LoG and Multi-level Thresholds

Abstract

An automatic defect detection system for the micro lens of a cellular phone was studied. We presented a defect detection method using LoG filter and multiplelevel thresholds. We could find general defects such as scratch and even the stain defects that the detection might be difficult because of the low contrast between its background and a defect in the lens images. We did experiment using 50 spherical lens images with defects. As the experimental results, it showed the average recall and precision are about 92%, 77%, respectively in the case of sigma=2.5.Sixth International Conference on Advanced Language Processing and Web Information Technology 0-7695-2930-5/07 $25.00

Optical Design of a Novel Collimator System with a Variable Virtual-Object Distance for an Inspection Instrument of Mobile Phone Camera Optics

“…The proposed inspection processes that use a specialized image-processing algorithm were designed to reduce the overall assembly time and defects for compact camera module assembly. In addition, a real-time noncontact eccentric-error-measurement system that uses an optical setup without a rotating and clamping mechanism [7] and a method for detecting foreign substances on the lens surface were also proposed [8,9]. The system could assess the 2D surface defects and non-uniformity of the 3D film thickness of specific materials using pattern recognition and morphological techniques.…”