Image Recognition Algorithm In Chinese Paper Cut Application

With the growing development of technology between Internet and mobile device,traditional Chinese arts and crafts have been seriously impacted by the scinece and technology industry,and many tradition arts is facing with the problem of no inheritance,especially paper-cut culture.According to the features and characteristics of Chinese papercut,this paper first presents a paper-cut processing algorithm which can transform the real image into paper-cut effect.It can convert any real images into paper-cut style,and we developed a mobile software on i OS and Android mobile platform.At the same time,we hope that the software developed in this paper can unite the current technology and traditional culture,so that users can feel the art of paper-cut anytime and anywhere,and make use of social network communication to contribute to the promotion of Chinese traditional culture.Traditional Chinese paper-cut is very sensitive to contour edge noise,and the image processing algorithm aslo requires high computation efficiency as for the limitation of the mobile device computing capability.The above two features,are the important challenge in the image processing technology of the paper-cut effect in mobile devices.By analyzing and summarizing the characteristics of traditional paper-cut,the most critical step in refining image and converting it into paper-cut works is to deal with the edge of target image.Therefore,we combine with a variety of target segmentation algorithms to preprocess the image in the paper-cut effect processing algorithm,then detect the image edge segmentation to reduce the impact of edge noise through a variety of edge detection operators,and extract the contour edge of the target image.As the same time,we provide flexible and artificial interactive contour optimization operation processing to further filter edge noise effect in order to achieve realistic paper-cut effects.In the application development of this thesis,as for the landscape image,the improved graph cut algorithm,called Grab Cut,is applied to preprocess the image based on the semi-manual interaction.Semi-manual interaction is not only a common way to deal with user interaction used in the mobile device,but also has high computing efficiency and stable precision of the segmentation.In addition,image segmentation algorithm can provide an accurate for further contour extraction and eliminate unwanted image backgrounds.As for special image with people,this thesis uses the special face recognition technique to correct the effect of different illumination,uses a large number of skin detection database to match the image to enhance the accuracy of the face edge detection,and obtains the portrait of the paper-cut image.At the same time,the special treatment of portrait in the algorithm also make up the shortcoming of two traditional forms of expression in the face and other images for more details,and solve the problem of undesirability segmenting a particular image for transferring paper-cut to the online experience.Secondly,we propose an optimized anti-noise operator to detect the edge of object,by analyzing the advantage of the first derivative operator and second derivative zero cross operator.In order to make the edge of extracting contour more clear and significant,we further apply the morphological open operation for the extracted edge to optimize the contour,which can help to eliminate noisy point and fill the cracked local infrastructure,in order to get the better image in paper cutting style.Finally,after the system architecture design,we implement the novel algorithm on both i OS and Android operation system,and construct an image-processing application for users to transform the conventional images into paper-cut images.What’s more,according to theoretical analysis and performance study of the sample test,this thesis validates the significant feasibility of the proposed scheme.