Research On Detection Method Of Automobile Dashboard Based On Computer Vision

With the development of the automobile industry,especially the extensive application of automotive electronics technology in automobile dashboards,the existing automobile dashboard detection algorithm has been unable to meet the detection requirements,and when it is actually applied to the automobile dashboards detection task,it shows many problems such as low detection accuracy and poor versatility.Therefore,the automobile dashboard manufacturers have an urgent need to construct an automatic detection system with high reliability,high detection accuracy and versatility.This paper aims at enhancing the accuracy,versatility and intelligence of the automobile dashboard detection algorithm.Based on computer vision technology,a new automatic detection algorithm is proposed.The original image has the phenomenon of uneven light and reflected light.In order to solve this problem,this paper proposes an external illumination processing method based on dark channels,which makes the light of the dashboard image more uniform and realizes the automobile dashboard detection under external illumination.The traditional algorithms need to set a lot of parameters in order to detect other automotive dashboards.For solving this problem,this paper proposes a method of extracting speedometers,tachometers,pointer area and digital characters based on object detection.This method constructs a special SSD model for automobile dashboard detection,which realizes the automatic extraction of the test areas.This design framework enhances the versatility and intelligence of the detection algorithm.This paper analyzes why the automobile dashboard pointer extraction algorithm has low precision and poor stability,and then proposes the pointer fine extraction algorithm,which provides high-precision pointer fitting results for automobile dashboard detection.In order to solve the problem that the traditional detection algorithm requires manual assignment for the extraction of tick marks.This paper proposes a method for extracting tick marks by region search based on character recognition.This paper introduces the digital recognition method based on character segmentation and ensemble learning.Separating connected characters by inertial drop fall algorithm.The ensemble learning classifier is built by combining the SVM character classifier with the CNN character classifier to achieve high accuracy and low false detection of instrument digital character recognition,and then automatically extracts the tick marks based on the region search.This brand-new method enhances the automation of the dashboard detection.The traditional detection algorithm simply takes the pointer rotation angle and the indication value as a linear relationship,which makes the pointer reading error larger.To solve this problem,this paper proposes a method based on Lagrange interpolation to describe this relationship.This design rule has high precision,simple calculation and can be applied to different automotive dashboards.The experimental results show that the automotive dashboard detection algorithm based on computer vision has good detection accuracy.Compared with the traditional detection algorithms,this algorithm has good generality and automation,can meet the actual automotive dashboard detection needs,which has broad application prospects and practical value.