Comparative Study of Three Machine Visual Recognition Algorithms in Identification of Un-vibrated Concrete

Mechanical automatic vibration of concrete is a development direction of civil engineering construction, and automatic vibration accuracy control algorithm is one of the difficulties. This article attempts to use machine vision solutions to solve this problem. Program for three common feature extraction algorithms (Hog algorithm, LBP algorithm, Haar algorithm) is written through Python language. Then the “Support Vector Machine” algorithm is applied to classify the feature data extracted by each algorithm. By this way, the accuracy and efficiency of the three algorithms under different parameters are compared. The test comparison results show that the Hog algorithm is more efficient but the accuracy is average, the LBP algorithm has the highest accuracy but the efficiency is general when multiple training is required, and the Haar algorithm has the relatively lowest accuracy and efficiency. The results can provide a basis for the selection of different algorithms and parameters for various scenarios where the visual inspection and recognition system of un-vibrated concrete may be applied.