Técnicas de segmentação de imagens na geração de programas para máquinas de comando numérico

Abstract

This work describes a tool for extraction of features in 2D images applied to industrial automation. The method uses techniques implemented targeting of images into pieces model as detection of edges in the extraction of spatial coordinates of the image and import applications in CAD (Computer-Aided Design) for the generation of programmers for Computer Numerical Control (CNC). This work is to copy a particular model (image) from an existing object, which features a process of reverse engineering. An initial prototype was developed using the Java language with the Eclipse IDE and library for handling complex images, the API Java Advanced Image (JAI). For the acquisition of parts to be machined, was prepared an environment of lighting to highlight areas of interest improving the luminance of the same and enabling the application of techniques developed in this prototype. Different operators for detecting edges were exploited in order to implement the best technical and that best represent the real piece. A new technique of targeting the prototype was adapted from the integration of technology Java / MatLab and with the acquisition of images through technical Multi-flash. After the application of different processing techniques, a text file containing the coordinates of the image (piece) is generated and imported into CAD. In the CAD environment, runs calculating the trajectory of the tool that identifies the geometry of the piece and defines the path of the tool in the generation of CNC program for breeding the same. Results presented and evaluation of the tool demonstrates the feasibility of applying the automated system developed as part of parts to play in numerical control machines.