The Research On The Work Site Steel Pipes Counting Based On The Digital Image Processing

Under the traditional on-site environment, steel pipe counting has been using the manual counting method, which inefficient, and lower accuracy rate which have led by easy fatigue and other human factors. Content to free from the complicated counting labor, this paper combined with the existing research, aimed at the infeasibility of the current method for the Work site Counting, presents a new method based on sample set extracting. The paper introduces the key technology and related algorithm. The method finds the sample set by the way of holes filling in the morphological reconstruction, matches and statistical counting with the gradient image by the sample set. It achieves the steel pipe counting in the pipe processing and movements, and has industrial applicability.The proposed method consists of image preprocessing, edge extraction, the sample set extraction, templates, production, testing counting steps. Firstly, convert the collected image to gray-scale images, then the the GRA method and Top-hat method joint enhanced the image to improve image quality. GRA method based on grey theory, to extract small sample reference sequence according to the weak correlation of the noise point with their field pixels and the strong correlation of edge points with the pixels in its field, to analysis of small sample data associated with the Deng’s incidence degree model, to identify edge point and non-edge points, and enhance the useful information. Top-hat algorithm combined with the subtraction, and the opening and closing operation gray-scale morphology for enhanced bright objects on a dark background image. Through the combination of the two above, for the harsh environment of the Work site, the enhancement effect significantly. Then extract the edge from the improved image, and extract the edge of the complete object, i.e. the sample set, with the hole filling method. The Hole Filling is based on morphological reconstruction method. With the hole filling, filled areas of the image can be obtained, then the paper get the edges of objects in the image through the edge reconstruction of the hole filled areas.aggregate and weight the extracted edges of objects to form the template of the mask weights. Mark the feature of the mask template and match and detect the object edge of the entire image with the weighted template, and ultimately statistics count the result of the detection. It has good applicability of the method for the multi-object counting and holes filling is a simple method to find the sample set compared with other multi-object recognition and the effect is significant.Finally, through MATLAB programming, validation experiments prove that the method of image enhancement effect is significant, the final count of good results,It can replace the human hand-count, the promotion and application value.