Diesel vehicle tail gas smoke intensity detection method based on deep residual error learning network

Abstract

The invention discloses a diesel vehicle tail gas smoke intensity detection method based on a deep residual error learning network. The diesel vehicle tail gas smoke intensity detection method comprises the following steps: acquiring a diesel vehicle tail gas image by using a camera mounted on a road; acquiring a label value of the deep residual error learning network by using an integral gasoline-diesel oil automobile tail gas telemetering device; establishing the deep residual error learning network; pretreating the acquired diesel vehicle tail gas image, training, verifying and testing the established diesel vehicle tail gas image, and applying the obtained deep residual error learning network to real-time on-line detection on the smoke intensity of a diesel vehicle tail gas. Compared with a conventional method, the diesel vehicle tail gas smoke intensity detection method has the advantages that the depth of a network can be increased on premise that the performance of the network is ensured, and a system has relatively high accuracy and relatively good generalization, so that the diesel vehicle tail gas smoke intensity detection accuracy can be improved.

Description

technical field [0001] The invention belongs to the technical field of image processing, and in particular relates to a diesel vehicle exhaust smoke detection method based on a deep residual learning network. Background technique [0002] With the advancement of China's urbanization, the number of motor vehicles is increasing. At the same time, the pollution of motor vehicle exhaust to urban air is also more serious, so it is urgent to control the pollution of motor vehicle exhaust. The premise of governance is to have a full understanding of the emission of motor vehicle exhaust, so it is necessary to need a fast and accurate detection technology for motor vehicle exhaust. Diesel engine has high power and good economic performance, which is an indispensable driving force in urban construction, and a large number of diesel vehicle exhaust pollution follows. Heavy-duty diesel vehicles are an important contributor to air pollution because the amount of nitrogen oxides and par...

AI Technical Summary

Problems solved by technology

Its obvious disadvantages are: it needs to perform a series of operations on the target area in the exhaust image, and the steps are relatively cumbersome; more importantly, the accuracy of each step of operation (such as threshold segmentation) will be affected by the surrounding environment (such as light, fog, etc.) Haze), resulting in inaccurate detection or even failure

[0006] As far as the selection of the number of network layers is concerned, 1 to 100 layers are all possible, but for the detection of diesel vehicle exhaust smoke, the number of layers is too small to make the accuracy of the deep residual learning network not high enough, and the number of layers is too large for training. A lot of cost is required in the process

Images