Research On Image Layout Extraction Algorithm Based On Hadoop

With the rapid development of integrated circuit industry, ultra large scale integrated circuit came into being. In the manufacturing industry of integrated circuit, the finished product rate of integrated circuit has been the concern of the manufacturing industry. The optimization of the layout of the production process to improve the finished product rate of the integrated circuit has become a hot research. In the image layout, the layout of random defects is more intuitive, so the layout of the image from the CIF file is a necessary choice to solve the problem. Hadoop technology is usually used for distributed processing of large amounts of data. Because of the large amount of data, the large scale of CIF map is a long time, Hadoop can be used to distribute these data. So it can be used to improve the extraction efficiency.In this paper, a Hadoop based image layout extraction system is designed, which includes the front-end control system and data storage and processing system. The front control system is responsible for uploading the CIF file, editing the distributed processing program and controlling the running of the system. Data storage and processing system to complete the CIF file of the sub block storage and distribution of the layout of the distribution. And then, an algorithm for converting the file to the format of the CIF file is proposed. The algorithm is applied to the layout of the Hadoop distributed processing platform. Firstly, the algorithm of converting CIF file to BMP image is designed by using Map Reduce programming idea. Then, based on the Hadoop platform, the image elements of the large scale layout are extracted from the Hadoop cluster, and the data is distributed by the distributed processing system HDFS. Image layout extraction algorithm based on Hadoop, using HDFS distributed processing system processing data, compared with the existing single machine based image layout extraction algorithm can improve the processing efficiency of data. In addition, the pixels of the image data in the algorithm are not stored in RGB format, and only one bit of memory space is saved, so the image layout can be improved.