Preliminary image processing in vision systems

Annotation: Currently, cameras that use a matrix of photosensitive elements are widely used for video. Matrixes based on chargecoupled devices or arrays of photodiodes are used as image sensors in such cameras. These cameras allow you to get an image of the external environment, divided into three color components – red, green and blue. To reduce the resource requirements for the stage of image processing of the working scene of an industrial robot, pre-processing of the image obtained from the camera is performed. This stage involves the transformation of the original image into a halftone and the transformation of a halftone image into a binary one. A large number of methods have now been developed for converting a halftone image into a binary one. The authors of these methods are Otsu, Gonzalez, Woods, Eikvil, Bradley, Roth, Bernsen, Niblack, Sauvola, Singh, Wolf. The binarization methods proposed by these authors have different properties. Among these methods it is important to highlight the methods that preserve the most important characteristics of the objects of the working scene of an industrial robot. The analysis of the most well-known methods of binarization.

Keywords: industrial robot, technical vision, image processing, binary image