![]() ![]() Finally, the generated background is statistically evaluated, and segmentation result of the dirt particles is used to estimate the quality of the semisynthetic images.įine and sparse details appear in many quality inspection applications requiring machine vision. It is shown how the background for dirt particle images is generated, what is the principle of dirt particle placing over the background, and how the problem of color normalization is solved. This paper introduces the algorithm for the purpose. In order to combine dirt particles of different types in one image of a paper sheet and to know the exact location and the type of the particles, a semisynthetic method for generating the ground truth was developed. Consequently, the initial data was provided as a set of paper sheets with a singl dirt type in each. In the present work, laboratory personnel produced pulp and dirt particles for the samples as pure as possible. To get suitable samples of paper with expert-annotated dirt particles in each can be considered as a too complicated, laborious, and time-consuming task. For such comprehensive evaluations, reliable ground truth is essential. Automatic classification methods can be designed for the task, but there should also exist proper evaluation data to truthfully compare the methods. In the evaluation of dirt inclusions in paper, the attention is paid not only to the quantity of dirt but also to the type of dirt particles. Multiresolution/multiscale models that aim at representing highly complicated random fields include various nonlinear hierarchical/multiresolution/ multiscale texture models. Through their connection to multigrid methods, these models often can improve convergence in iterative procedures. Multiresolution/multiscale models that aim at computation reduction include various multiresolution/multiscale MRFs and multiscale tree models. Specifically, multiresolution/multiscale processing can provide drastic computation reduction and represent a highly complicated model by a set of simpler models. This interest has been motivated by the significant advantages they may have in computational power and representational power over the single-resolution/single-scale models. With the advent of multiresolution processing techniques, such as the pyramid and wavelets, much of the current research in random field models focuses on multiscale models. It also discusses the multiscale tree model and a model based on the Gaussian mixture. This chapter provides an overview of random field models, emphasizing the autoregressive and Markov fields. Using the standard classification methods, it is possible to obtain satisfactory results, although the methods modeling the data, such as the Bayesian classifier using the Gaussian Mixture Model, show better performance. The feature selection proved to be important when the number of dirt classes changes since it allows to improve the classification results. The results of the experiments show that the semisynthetic procedure does not significantly change the properties of images and has little effect on the particle segmentation. The framework was tested both with semisynthetically generated images based on real pulp sheets and with independent original real pulp sheets without any generation. Sequential feature selection was employed to determine a close-to-optimal set of features to be used in classification. To classify the dirt particles, a set of features were computed for each image segment. #Pixel 3 dirt 3 backgrounds manualTo avoid manual annotation, dry pulp sheets with a single dirt type in each were exploited to generate semisynthetic images with the ground truth information. ![]() ![]() This paper proposes a framework for developing and tuning dirt particle detection and classification systems. However, the ground truth required to train an automated system is difficult to ascertain, since all of the dirt particles should be manually segmented and classified based on image information. Since manual quality control is a time-consuming and laborious task, the problem calls for an automated solution using machine vision techniques. Knowing the number and types of dirt particles present in pulp is useful for detecting problems in the production process as early as possible and for fixing them. One important aspect of assessing the quality in pulp and papermaking is dirt particle counting and classification. ![]()
0 Comments
Leave a Reply. |
Details
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |