The Study On Algorithm And System Implementation Of Printing Image Online Detection

To increase the automation level of China printing industry and to aimed at the serious lag of theories study and system development to online quality detection of high-speed printing image, real-time processing algorithm and theory of motion image that based on machine vision technology is studied in the thesis, and some key technologies of high-speed presswork quality online detection system are realized.Firstly, a distributed detection system scheme based on Server/Client mode and an image data transmission method based on TCP/IP protocol are proposed according the analyses to the technical requirement and application features of presswork detection, and which solved the problerm of multi-cameras image processing and image data transmission effectively.To guarantee a uniform processing benchmark to the grabbed printing image, a judgment criterion to grabbed original image quality is set up. A gray processing method of true-color image that based on the color sampling is proposed, and a piecewise linear transformation is designed to enhance the contrast of gray-scale image. Presented a directional complement algorithm that base on the mathematic morphology to eliminate motion blur in note figures printing image, and an optimum gray value thresholding processing method based on iterative operation is studied.Through the analysis of major factors that influenced the registration of presswork image, a rapid extraction method based on gray value morphology operation is put forward to extraction the edge features of presswork image. To rapid and unified registration the features include marking, gray value distribution and linear features, a small neighborhood search registration algorithm based on feature subspace is put forward. An image match algorithm based on dynamic reference and association of positive cost and negative cost is put forward to realize the concurrent extraction of light and dark defection.A region segmentation algorithm that based on projection histogram is adopted to note figures division. A variable thresholding method and minimum distance criterion based state space search strategy is adopted to defection extracting and clusters merging. A modified NN BP algorithm is designed and applied on recognition of printing note figures and classification of printing defections successfully.At last, a fluorescent illuminating system and a four-freedoms adjustable camera clamp are designed to fulfill the requirment of image collection. Printing image online detection system and detection software that applied in cigarette packing detection and note figures detection are realized, and the detection systems have been successfully applied in printery.