Full grading detection method of earth-rock materials by combining preliminary screening and video image recognition

for detecting gradation in earth-rock materials using images face challenges, including difficulties in identifying fine-grained stone materials and significant occlusion between different earth-rock components. The present study proposes a comprehensive grading detection method for earth-rock materials that integrates preliminary screening with video image recognition. Initially, a preliminary screening-shooting integrated detection device is developed based on the particle size range and the proportion of fully graded earth-rock materials. The device is capable of screening fine particles measuring less than 5 mm and capturing video images of coarse-grained stone with significant separation on the slope surface. Second, an extraction model for the effective frame is constructed, along with a mathematical model for full gradation conversion that takes into account the slip velocity of earth-rock particles, based on the image preprocessing and particle contour feature extraction method for earth-rock materials. Finally, an accuracy analysis model for earth-rock grading detection has been established, leading to the development of a comprehensive grading detection system. This system effectively performs full grading detection, generates grading curves, and conducts accuracy analysis of earth-rock materials. The test results demonstrate that the method outlined in this paper is applicable for conducting six groups of earth-rock grading tests. The results of the grading test align well with the sample data, demonstrating an accuracy exceeding 97% and a relative error maintained within 2%. This research addresses the challenges associated with identifying fine particles smaller than 5 mm and the occlusion issues between earth-rock materials during the grading detection process using image recognition. This study establishes a theoretical basis for subsequent investigations into the applications of complex construction sites, facilitating the swift and precise identification of the complete gradation of earth-rock materials within these environments. ● A frame extraction model considering particle sliding velocity is constructed. ● Automatic classification detection of earth-rock materials is realized. ● An earth-rock grading intelligent detection system is developed.