| Synthetic images,a sort of widespread digital media nowadays,are normally created and synthesized to meet the demands of certain applications.Consequently,there exists potential regularity that reflect the creation intention of the designers in the content and structure of the images.Nowadays,printing and dyeing industry is a representative application field of wide-ranging types of synthetic images.Due to the high design cost of printing patterns,which are formulated as vector images and designed by professional artists,traditional printing industry has trouble adapting to personalized printing situations which is influenced by the developing of digital printing techniques Finding repetitive patterns that embody the creation intention of designers in synthetic images,and using the repetitive pattern instances to synthesis new printing patterns,can help to analysis the content of the images effectively and assist the printing pattern creation process of users,which provide a new train of thought of personalized textile printingsThis paper takes printing images as a representing of synthetic images,and analysis their fundamental features.On this basis,we research on the methods of discovering repetitive patterns and synthesizing printing patterns with various types of synthetic images.Generally,the main contributions of this paper are:(1)Presenting and implementing a method of repetitive pattern discovery with non-overlapping objects.Based on the definitions,we formulate a unified representation of repetitive patterns,and present a method that utilizes content uniformity and spatial structure uniformity of objects simultaneously.The results show that the method can deal with non-overlapping objects with multiple categories and scales,and discover the pattern scale and instances in the certain images effectively and precisely.(2)Presenting and implementing a method of repetitive pattern discovery with overlapping objects.We present a method that utilizes the repetitive spatial constellation of objects to discover the pattern scale,and further using the scale to conduct the detection of object instances and their categories,and discover the instances of repetitive patterns afterwards.The results show that the method relieve the object detection errors caused by the overlapping states between adjacent objects,and discover the pattern scale and instances in the certain images precisely.(3)Presenting and implementing a method of synthesizing printing patterns based on discrete water-tight layouts.We present a method that employs brief user-interaction to assist users-input parameters,and utilizes the integer liner programming to solve the optimal discrete water-tight layouts,and finally renders the printing patterns in the canvas space.The results show that the method can meet the demands of continuity and homogeneity of printing patterns,and achieve the goal of synthesizing user-defined printing patterns with low user-interaction cost.The method provides a new train of thought of personalized textile printings successfully. |
