A Method Of Extracting Movable Parts And Generating Animation For Sketch Based On Big Categories Classification

Sketch animation can show a rich story through the movement of simple lines.In traditional methods,sketch animation is often produced by frame-by-frame rendering,which has a high labor cost and can not be generated in batches.Many simple brush objects have similar characteristics,such as birds with wings,bipedal animals with two legs,and so on.When designing animation,they perform similar actions.This paper presents an automatic method for generating sketch animation.This method can automatically generate animation for user input simple strokes,save the time of manual drawing.At the same time,it can be combined with the sketch animation story system developed by our research team.It can automatically generate animation deduction stories for children's paintings and inspire children's enthusiasm for learning and exploring.The method is divided into three steps: recognizing the motion category of the object,extracting the moving part strokes,and transforming the moving part strokes to generate animation.Taking stick figure with fixed reference template as the research object,this paper designs a classifier Trans-Net for the classification and recognition of stick figure object,and introduces transfer learning and data enhancement to solve the problem of small data set.According to the motion characteristics of simple strokes object,this paper divides four kinds of simple strokes into four categories,and designs a method of extracting movable parts and generating action for them.Through the relationship between the type and position of strokes,we can judge whether they belong to the specified movable parts,and generate simple animation by translating,rotating and scaling the combination of strokes belonging to the movable parts.Experiments show that the method can be applied to all types of simple stroke objects in each large class.