Resrarch On Rmb Image Enhancement Compression Technology

RMB is closely related to people’s life.There are only six banknote printing factories now,which are distributed in six regions of China.The Central Bank not only supervises the output of RMB but also the quality of of RMB.Under double pressures of yield and quality,strict quality detection and promotion of detection speed are vital,one banknote printing factory bears the hundreds of millions of money to detect one day.In order to cost decreasing and benefit increasing,false RMB images detected by the machine must sentenced to manual online detection program,so data of manual online detection program can improve the machine.So the key point of this paper is the link between machine and manual work.Purpose of this paper is to use the highest compression ratio under visual lossless standard to storage bad RMB images of machine,and transfer bad images to the manual online detection program as fast as possible.At first,this paper compares three mainstream compression standards:HEVC(High Efficiency Video Coding),JPEG2000 and JPEG-LS(ISO/ITU-T-14495-1,87)from compression speed and compression ratio.Then the paper chooses JPEG-LS compression standard to improve because of its low complexity.So that the transmission time and compression space of RMB can be saved and continuous production can be guaranteed.In this paper,two methods are proposed to improve the compression ratio of JPEG-LS.The first method is to improve the compression ratio through RMB characteristic cutting.Without changing the encoder,the 100 RMB image is cutted according to the anti-counterfeiting features and distortion zone features.The image after cutting according to the anti-counterfeiting features is compressed in lossless way,and the compression ratio after cutting is raised from 1.79 to 1.81.In order to achieve the highest compression ratio on the basis of visual lossless,different distortion quantization parameters are selected for near-lossless compression of different parts of the image after cutting according to the distortion zone features.The compression ratio obtained by this cutting method is nearly two times higher than original lossless compression.The second method is to change the trait of JPEG-LS to compress image as a whole,and set distortion quantization parameters of three different RGB channels separately.According to the color characters of new 100 RMB and 10 RMB,the RGB channels can use different distortion quantitative parameters to compress.The author decided to set a higher distortion quantization parameter for red channel of 100 RMB,and set lower distortion of quantitative parameters for the other two channels.And set a higher distortion quantization parameter for blue channel of 10 RMB,and set lower distortion of quantitative parameters for the other two channels.Finally the optimal distortion quantization parameters of three color channels of 100 RMB are 8,3,4 and the optimal distortion quantization parameters of three color channels of 10 RMB are 4,5,9.The compression ratio of this method is nearly two times higher than original lossless compression.