Research On Image Acquisition And Recognition For Raised Characters Based On Structured-Light

Industrial tags are widely used in mechanical and electrical products. The characters carrying product information are formed on the surface of objects by machining method. These characters are called raised characters because they are three-dimensional characters and they have the same color as the background. It is this characteristic that makes it difficult to use the existing research achievements of the traditional Optical Character Recognition (OCR) to recognize them. Recently, with the rapid development of the Manufacture Information Engineering (MIE), auto-recognition system for raised characters becomes a necessary requirement for enterprise modernization. Supported by the Young and Middle-Aged Scientists Research Awards Fund of Shandong Province, the research carried out in this dissertation is as follows.Traditional optical three-dimensional measurement method is appropriately improved based on the questions which this article will solve. The main point of the system is focused on the three-dimensional characteristics of characters. Firstly, according to the principle of active triangle measuring, functional relationship between the image coordinate system and the world coordinate system is established. The theoretical foundation of the image acquisition system for raised characters is set up.Based on the parameters of the characters to be measured, the hardware that system uses is decided. And as a technical support, the key parameters of the system are calibrated. Furthermore, the theoretically-calibrated parameters are verified, in consideration of the small size of characters and the precision of system. An experimental rig is established and the preliminary experiment is conducted. To eliminate the blocking defect of this system, the author makes a further improvement of the calibrated parameters, which makes the experimental result basically meets the requirements.Then the experiment rig is revised and a large number of experiments are conducted. The images obtained by experiments are then processed, and the threshold value is appropriately determined on the basis of the analysis on the distribution pattern of gray values of these images. Then the threshold gray value transform is applied to the original images, and after filtering the background, the centerline of the structured-light is abstracted. An improved pruning algorithm is applied to eliminate the burrs and guarantee the continuity of the obtained centerline. For those images with defects due to the uneven reflection of light of the tag material, corresponding image enhancement methods are used to enhance the useful information such as the distortion of structured-light. A gray value transform algorithm of the height information of the characters is presented. This new image processing algorithm transforms the distortion degree of the centerline of structured-light into corresponding gray value, which simplifies the complex algorithms in optical measurement and achieves the design goals of this system. The results of image addition completely represent the contours of characters. At last, through comparing with gray images obtained by other methods, the feasibility of this system is verified.The characteristic of character recognition for this system is studied. A more rational approach is developed for character segmentation and size normalization. According to the ultimate purpose of this system, character recognition is based on template matching and VTD-HTD feature recognition. The result of the experiment indicates that single identification algorithm can’t meet the requirements of the system, further more an advanced method of template matching based on the VTD-HTD feature is introduced. And the experimental results show that the recognition rate of the new method can reach to100%in the experimental condition of this thesis.