Interactive processing of ferrule images to automatically estimate pollution degree

The article is devoted to development of an processing interactive algorithm for ferrule connector image. This image is used for pollution degree estimation and insertion loss and reflection coefficient of fiber optic. One of the main problems in image processing is recognizing the contour of the “cladding layer-ferrule” boundary, which can be distorted by pollution. Currently this problem is solved by the operator. He manually sets the points on the desired contour, which are then used to approximate a circle. The stage of manually defining the points takes a lot of time and requires repetition if errors occur. The paper presents an algorithm for automatic recognition of the “cladding layer-ferrule” boundary contour, which simplifies the operator’s task. The interactive stage involves setting a rectangular area that approximately limits the desired contour by operator (installer of fiber optic). Experimental testing of the developed algorithm showed the quality of recognizing the “cladding layer-ferrule” boundary in images.