The Print-defect Detection Method Based On Artificial Immune Algorithm

In the printing industry, print-defect detection plays a major role in the assessment and control of print quality. Prints with defects are supposed to be eliminated in the actual production. However, at the present stage, the print-defect detection methods have more or less problems and need to be updated. The development of image processing technology and the emergence of new intelligent algorithms promote the progress of detection technology. The integrated application of various intelligent algorithm is expected to develop a new and more efficient print-defect detection method.This paper studies the print-defect detection method based on artificial immune algorithm.Artificial immune algorithm has the advantage of it can establish multi-layer detection mechanism, which need a little or no prior knowledge, and only need a few types of defect sample. Based on this, the print-defect detection method based on artificial immune algorithm has a broad development space.On the basis of studying on negative selection principle in artificial immune algorithm and image processing technique, Gray level co-occurrence matrix ( GLCM ), which can reflect distribution characteristics of pixel gray level, is employed to obtain texture feature values(Energy, Entropy, Contrast, Correlation, Homogeneity ) of print images in the respective direction of 0°,45°,90° and 135 ° .The averages of these feature values are taken as the final value marking image characteristics. The averages of each feature value composes a feature vector. The feature vector is used as the data of the negative selection algorithm.This paper studies the basic negative selection algorithm and modifies it with the real value encoding. The sample data is represented by multidimensional real value vector. The Euclidean distance is used to calculate the affinity between the two samples and to determine whether they match. Moreover, " vaccine"is introduced to modify the basic negative selection algorithm for the identification of print defects types. This method constructs multiple detector sets, uses multi-detector sets fusion detection method to diagnose defect type. In order to improve the detection accuracy of the algorithm and reduce the misjudgment rate, this paper proposes a method to match the sample with own to be tested twice in the detection phase.Finally, this paper realizes the whole process from image processing to defect detection,and designs the print-defect detection system based on MATLAB.The experimental results show that the modified negative selection algorithm can effectively detect the print-defect. The print-defect detection system responds rapidly in the identification of print defects and defect types with high accuracy. It has a certain application value.