The detection method of the silicon melting process in the CZ single-crystal furnace

The melting process is one of the main processes in the Czochralski method. Detecting the melting process is important for entire automation of the CZ crystal furnace. In regards to the furnace melting process, this paper presents a kind of image-based method to detect the size of the silicon ingot at later stages of the process. It could achieve the purpose of optimizing the heating power and process automation by monitoring the process. In the summary of the basic characteristic of the image of melting process, this paper proposes an image processing method which integrates silicon ingot color features, a multi-scale morphological gradient, an improved watershed transformation and the Bhattacharyya distance to separate the silicon ingots from the image background.