Article Preview
Top1. Introduction
DPM is the abbreviation of Direct Part Mark, it was originally used in machinery and electronics industry, and then extended to the field of mechanical manufacturing, pharmaceutical, medical treatment, military and other industries (Li Juan, Su Quan-qun & Liu Li-mei, 2009). One of the biggest features of industrial DPM code is permanence. When the parts are in negative or harsh environmental conditions, direct part mark can achieve permanent and whole life-cycle tracking ability. Commonly, the superficial area allowed to mark directly is relatively small, so DPM image must select high coding capacity codes. Data Matrix code has high coding capacity, high density, superior information security and other characteristics. When compared to other 2D bar code with the same size and density, DM code contains the maximum information (Li Jing-ya, 2009). So Data Matrix code has become the most frequently used type in DPM image. Due to the diversity generating methods and mental materials, collected industrial DPM bar code images usually exist uneven illumination, low contrast, much noise interference, complex background and other unfavorable conditions. How to identify it quickly and accurately is an issue. Also in some cases, background area color of collected images is darker than bar code region, while the existing 2D code recognition device can only recognize the case which bar code area color is darker than the background region, the invert-color processing is required. In addition, metal DPM codes can be divided into dot DM codes and standard codes, while existing code scanning device can only identify the standard type, it is still needed to find a method to standardize the dot data matrix images. Therefore, in order to identify such bar codes quickly and accurately, a preprocessing procedure with appropriate treatment effect, strong anti-interference ability and adaptive capacity is of particular importance.
In General, preprocessing procedure consists of the following steps:
Among them, binarization is the key point. In this paper, the preprocessing procedure is started with the characteristics of DPM 2D bar code, first use the homomorphic filtering method to weaken the effect of uneven illumination, enhance image contrast degree. After that, do horizontal and vertical projection, find the points with greater intensity changes in both directions, make the image into blocks, again use the classic Kittler binarization algorithm on each block, meanwhile, the mathematical morphology method is used to standardize the dot data matrix images. Finally, detect the “L” type finder pattern, find its pixel value, if the color of the background region is darker than the bar code area, do the invert-color processing, to insure the existing code scanning device can identify DPM code quickly (GAO Tao, DU Xiao-cheng, FISTER Jr. Iztok, 2014).
Top2. Homomorphic Filtering
For a gray image f(x,y), which can be represented as the multiplication of illumination and reflectance model. The amount of source illumination incident on the scene being viewed denoted as i(x,y), it is relatively uniform, with less space location changes, and occupies low-frequency part, which corresponds to the image background. The reflectance component of the objects on the scene denoted as r(x,y), it depends on the nature of the object itself, because of the nature and structural characteristics of different objects, reflected light intensity is different. With spatial position changing more severe, the reflected light occupies high-frequency part, which corresponds to image details (Ramaraj. M, S. Raghavan & Wahid A. Khan, 2013). Then the original input image f(x,y) is defined as: