Research On Weak Classifier Adaptive Integrated Enhancement In Image Recognition

Steel strip is one of the main products of iron and steel industry. With the changes of market demands, the throughput of high-quality steel strip is the most important factor when iron and steel enterprises compete with others in the international market. It has important theoretical and economic value that we research on the image recognition of surface defects in the steel strip.In this paper, based on the analysis of classified recognition ways that are used in surface inspection system of steel strip, such as the complexity and contradictions between the classification accuracy and classification algorithms, tradition classification methods have their drawbacks, and it’s hard to make new breakthroughs in the classification. So the new method of weak classifier adaptive integrated enhancement in image recognition was brought forward.The algorithm is that a number of weak classifiers are composed of many extracted simple features, and lots of weak classifier would form a strong classifier. The principle of this algorithm is a process of updating samples weight. Each value of weight indicates that the samples were misclassified cases. The value of weight will be increased by misclassified samples, and the next round of iteration will pay more attention to the misclassified samples of the last round of iteration. Weak classifiers will focus on the samples that are difficult to distinguish, by adaptively changing the distribution of the weights of training samples, so that it can reduce the error of classification, and greatly improve the effectiveness of feature classification.The advantage of this algorithm is that we can get a strong classifier, as long as a weak learning algorithm which is slightly better than random guessing can be found. And it balance the complexity and contradictions between the classification accuracy and classification algorithms very well.In this paper, we just use the above theory and method to classify six type defects on the surface of iron and steel strip:edge sawtooth, welding seam, mixed material, wrinkles, abrasion and slice. By experiments, we find that the theory can reach average recognition rate of94%and the recognition rate of board strips surface defects images can be improved better adjusted the recognition model.