The High-precision Detection System Research Based On The Granule Image

Performances and qualities of many products are significant impacted by granule size. Compares to traditional methods of granule size detection, image processing of computer vision technology is safer, faster and non-contact which becomes the develop trend of granule size detection technology.To achieve high-precision detection of granule image, a granule thickness measurement system is designed firstly. This dissertation is mainly focused on pre-processing algorithms of the uneven illumination, image segmentation based on watershed segmentation etc. The aim of this dissertation is to build hardware platform and design a software system which is based on image process algorithms to improve the efficiency and accuracy of granule measurement.Firstly, in order to avoid the negative environmental impact of the light, a novel granule image enhancement algorithm based on the wavelet transform is proposed, which combines histogram equalization and median filtering to enhance the quality of granule image.Secondly, in order to realize the segmentation of granule image, a new algorithm is presented. The novel algorithm combine the watershed algorithm based on morphological reconstruction, distance transform and region growth.Results show that the system could segment the granule image accurately and efficiently.Finally, the granule image is calibrated to achieve the size of granules and realize the measurement of granule size measurement system. And a distortion correction is introduced to analysis the result of the deviation. Experiment results show that the system could measure the particle image accurately, and realize high-precision measurement of particle size.