Pipeline segmentation design is the first step to design alignment sheet. In this step, several rectangular boxes are used to cover pipeline and each box will become the basic unit of alignment sheet design. After studying various pipeline alignment sheet mapping technologies, the author found that traditional manual design method, which can take advantage of designers’ subjectivity, causes low work efficiency. By reviewing and studying existing works at home and abroad, the author believed that it is possible and feasible to develop an automatic segmentation algorithm based on existing curve simplification algorithms to improve to improve the efficiency of pipeline section design and alignment sheet mapping. Based on several classical curve simplification algorithms, the author proposed the automatic segmentation algorithm, which automatically adjusts the location of rectangular boxes according to the number of pipeline/circle intersection points and pipeline/ rectangular box intersection points. Finally, through comparing time and result with the traditional manual method, the author proved the algorithm’s effectiveness and feasibility.
Automatic Segmentation Algorithm of Ultrasound Heart Image Based on Convolutional Neural Network and Image Saliency
