ABSTRACT
LED chip localisation is an important process during the testing and sorting of LED chips. Many template matching have been widely used for LED chip localisation. However, these methods require that the template image is artificially made, which may lead to misjudgement due to human fatigue. In order to overcome these shortcomings, we proposed an edge detection-based algorithm to detect the pose (position and orientation) of an LED chip automatically. The algorithm consists of three steps. Firstly, smooth the image histogram by Gaussian filter to segment the LED chip from background. Meanwhile, apply the image pyramid method to reduce computation. Secondly, according to morphology operator on the LED chip region to obtain the Region of Interest (ROI) of edge detect and detect the subpixel edge precisely. Finally, take the RANdom SAmple Consensus (RANSAC) algorithm as weighting factors to fit the subpixel edge and obtain the pose of LED chip. The results shows that the proposed approach successfully locates the pose of the LED chip. Personnel costs and misjudgement due to human fatigue can be reduced using the proposed approach.
