|
Machine vision performance testing and analysis of the metering device, requires a precise determination of the seed plate holes Seeding time series, for which a speed the process of detection method based on machine vision, involving the composition of the system equipment, rotating test detection, data acquisition , image processing, timing analysis, MATLAB program the detection error estimates, technical advantages and surviving defects. The results show that the machine vision CCD sensor used to detect the rotational speed of the process, in principle and practice is feasible and effective to achieve the speed of the process of detection of non-contact, no damage, pollution-free, high-precision, convenient, fast and automated. Speed ??machine vision processes detection principle, it is possible to predict that the identification of the mechanism of mechanical work, mechanical transient torque and power, with potential applications in the field of fault diagnosis of mechanical running. Blood samples were no special environment, lighting and technical performance requirements of the CCD camera, but the shooting interval size and uniformity of the seeding performance testing required to meet the accuracy requirements, selection 30Hz shooting frequency. Mechanically rotating position of rotation representative of the marker, the color requirements and the background has a large optical contrast, the shape of the larger rectangle of aspect ratio, size of 60 mm × 10mm can, in order to facilitate the imaging and subsequent image processing, speed detection result is not sensitive to the dimensions. The rotation marker and rotate objects adhesions consider convenience, no special requirements, and easy to fabricate. The design and implementation of the study was to detect and study the detection error, six speed tests for Metering Device. Possible speed range of the metering device test coverage, lower test environmental conditions require no special lighting. Removing the non-detection of the target process, the intake of image frame image rack, seeding housing removed from the remaining scene, from the size ratio and brightness will rotate markers prominent. Therefore, consider the image samples should be collected to avoid non-detection target. Gray frame image processing, image rotation processing markers belong to two regions with the rest of the scenery and background brightness on the gray level is characterized by a bimodal distribution and bottom lightness approximate zero, and that each of the two regions lightness mean difference is large, the standard deviation is small as well. The selection of the optimal distribution characteristics according to the color and brightness of the scene of gray-scale transformation algorithm model. Frame image binarization process, purports processed image leaving only the rotating markers (grayscale
|