The Realization Of Cartoon Image Superpixels Segmentation And Merging Algorithm

It will be well known that information technology promotes the development of society and slowly cover all aspects of society at the same time, which enables people to enjoy a more comfortable and convenient life, digital entertainment has become an indispensable part of modern people's life. Anime is a mainstream of the digital entertainment industry, with the development of China's population, demand for anime products is constantly improving. However, the design work of animation is very tedious, especially a large number of repetitive designs for a common content, which greatly increase the workload of the staff and affect the development of animation industry. This paper puts forward a solution to the problem of repetitive work in the process of drawing animation. The main contents consist of segmenting the cartoon images, segmenting images to each constituent element accurately, and storing and coding the results, it is provides a wealth of material for post-production cartoon.The selection of high efficient and accurate image segmentation technology is the key of the scheme. In this paper, after comparing the comprehensive performance of image segmentation algorithms, the SLIC(Simple Linear Iterative Clustering) superpixel segmentation algorithm is selected as the research content. Superpixel represents a sub-region of an input image that can keep consistent and local structure of the image, the superpixel segmentation is the process of aggregating pixels into superpixels. SLIC mainly converts the image to CIELAB color space and converts XY coordinates to the normalized five dimensional feature vector ( )TL, A,B,X,Y. Then we construct a metric standard for five dimensional feature vectors. Finally, we conduct local clustering of image pixels to generate compact superpixel block with the shape of similar size in order to complete the image segmentation. The algorithm is simple and easy to implement, and it needs few parameters. The generated superpixel blocks with good compactness and boundary, which are uniform in shape and size. Our algorithm is also better than the other superpixel segmentation algorithms in terms of the speed of operation and object contour.In this paper, we improve over segmentation phenomenon for the SLIC algorithm. Over segmentation phenomenon, namely, which utilize single pixel block to represent region that is divided into a variety of small superpixel blocks. There exists confusion between the connectivity of superpixel blocks. The innovation points we put forward are as follows:1. Test and enhancement of connectivity of superpixel segmentation algorithm. We use test connectivity to achieve the purpose of reducing the phenomenon of over segmentation in image segmentation, which provide the support to merge similar superpixel blocks.2. Propose the merging theory and realize the superpixel merging algorithm.3. The image color inspection mechanism is implemented to the segmentation result in order to detect the accuracy of image segmentation results and the integrity of edge contour segmentation.After implementing the above functions, we conduct performance analysis of the proposed algorithm. The experimental results basically meet the demand of the solution.