Font Size: a A A

Research On Two-dimensional Barcode Image Processing Algorithm And Its VLSI Design

Posted on:2014-02-23Degree:DoctorType:Dissertation
Country:ChinaCandidate:J H LiFull Text:PDF
GTID:1228330401467811Subject:Microelectronics and Solid State Electronics
Abstract/Summary:PDF Full Text Request
2-D barcode has been widely used in certificate data storage, logistics data storage,retails and documentation confidentiality because it has a wide range of advantagesincluding convenience, flexibility, low-cost, high-information, reliability andhigh-confidentiality. At present, image processing based2-D barcode identificationtechnology is not perfect, and there are two aspects which need improvements. At first,because of the influence of unbalanced illumination, geometry distortion and defocusingblurring,2D barcode image processing algorithm needs to be optimized and improved;the second point is that the barcode decode devices are almost based on ARM, DSP orother controllers, and these devices have disadvantages including high-cost, high powerdissipation and low speed. Using2-D barcode image processing application-specificintegrated circuits can avoid these disadvantages. Based on this background, this paperconducted an intensive study on2-D barcode image processing algorithm and its ASICdesign. This paper includes following aspects.1. An illumination equalization algorithm is proposed. In the acquisition process ofa2-D barcode image, non-uniform illumination exits. This can bring great difficulty to2-D barcode identification. In order to increase the identification rate, spec ial methodmust be used to eliminate this non-uniform illumination. In this paper, four structureelements with different scales are used to perform WTH (white top-hat) transformationand the four processing results are merged together. The experimental results show thatthis algorithm can eliminate the non-uniform illumination and protect image details.2. An edge protection filter algorithm is proposed. Barcode image acquired bycamera is often influenced by discrete pulses and impulse noises. In order to reduce thenoise influence, a self-adaptive filter algorithm is proposed in this paper. This algorithmfirst judge if there is a median exits in the filter window. If a median exists, and themedian is not a noise, then the median would be output; if the median is a noise, theneight sub-windows including the median with different edge directions would beconstructed to detect the median of filter window, and the sub-window median with theminimum discrete degree would be the final output. If the filter window has no median,the eight sub-windows with different edge directions would be constructed and choose the sub-window median with the minimum discrete degree as the output. Theexperimental results show that this method can filter2-D barcode noises whileprotecting the image details.3. A2-D barcode location algorithm is proposed. Barcode location is a problemwhich must be solved in2-D barcode identification. The proposed filter algorithm firsterode the original image with size gradually increased structure elements, and thenacquire a sub-graphic whose connectivity is larger than a certain scale throughmathematic morphology reconstruction. Then this sub-graphic is compared with the lastone, when these two sub-graphics are the same, the acquired sub-graphic is the roughsegmentation result. At last, in order to eliminate the influence of noises, openingreconstruction is performed on the rough extracted image and acquires the fine extractedimage. The experimental results show that this method can locate different kinds of2-Dbarcode with different scales.4. An image correction algorithm is proposed. Because of the geometric errors andhuman made errors, linear and non-linear distortions exist during the lens transitionfrom3-D scenario to2-D scenario and these distortions can do great harm to theaccuracy of2-D barcode correction. In order to acquire the true information included in2-D barcode, distorted image must be recovered. This paper proposed a method whichperforms scan and fitting in turn, and deletes points with high discrete degree in severaltimes. This method can obtain an accurate edge line and furthermore obtain the accurateimage correction points. Meanwhile, to reduce the algorithm computation, the proposed2-D barcode correction algorithm performs inverse perspective transformation first, andthe affine transformation second.5. A configurable2-D barcode image processing integrated circuit is designed. Atfirst,2-D barcode image processing procedures including gray transformation,illumination equalization, noises filtering; binarization, location and correction areimplemented using Verilog HDL. These components share a SRAM memory and areconnected and compose a reconfigurable circuit. Secondly, this circuit is implementedon FPGA to perform function verification, the experimental results show that this circuitcomplies with the design requirements and achieves a good proce ssing result. At last,based on the function verification, this design is implemented using SMIC0.13μmCMOS process.
Keywords/Search Tags:Two-dimensional bar code, light balancing, Deformation correction, location, Configurable
PDF Full Text Request
Related items