Research On Missing Tooth Detection Algorithm Of Electric Shovel Based On Machine Vision

The electric shovel is an important mining equipment widely used in the mining industry to dig and load ore material,during operation,because of the impact of the ore material’s reaction force,the tooth in the bucket can be lost easily.The missing tooth will be mixed with the ore material and picked up by the haul truck,and then they will be send into the crusher.The crusher can be jammed and damaged as the result of the hard tooth’s force.It will lead to huge economic losses if this happen.So in this paper,we use the infrared machine vision system to monitor the status of the electric shovel bucket tooth,and combined with the image processing technology to study the detecting algorithm to detect whether the tooth are lost.The main content are as follows:Through the analysis of the electric shovel’s working process we can know that the shovel tooth will irradiate infrared beam as the result of continuous friction with the ore.Based on this,we take an infrared thermal camera to monitor the shovel tooth.So it can acquire high quality monitoring images,and at the same time meet the electric shovel’s working requirement of day and night non-stop.About the detection algorithm,first,the object detection framework of using the histogram of oriented gradients(HOG)feature and the support vector machine(SVM)algorithm is used to detect the teeth object in the infrared image.Second,based on this traditional object detection method,in this paper we analyzes the similarity relationship of the shape feature between the bucket teeth,and then we use the shape-context algorithm to extract shovel teeth’s shape feature,and use the shape matching method to add a shape feature constraint during the object detection process to improve the accuracy of the object detection.After accurately detecting the teeth object in the image,through the analysis the relative motion relationship between the shovel bucket and the infrared camera,we find the invariance of the spatial position of the bucket teeth in the infrared image,so we can use this to determine whether a teeth is lost.Finally,through the experiment,the whole detection algorithm is verified and the reason of the error is analyzed,and some improvement ideas and methods are put forward.