The Research On The Examination System In The Automotive Dashboard Based On Machine Vision

As one of the important parts in the car, automotive dashboard plays a necessary role among all the parts. It is in charge of recording all kinds of car's data in time, and responding the status of car, it looks like a window between driver and car. Currently the general quality test of such instruments is artificial observations to determine the quality of products, which asks for observing the situation between the pointer and instrument calibration. But this approach can cause low efficiency and large errors due to human factors such as the angle of observations, observation distance and the level of fatigue and so on.The study on the intelligent inspection methodology thus has great theoretical and practical value. In order to improve efficiency and accuracy, using an efficient, reliable method to replace the artificial detections has become a necessary trend. Therefore, this paper proposes a method which adopts Hough Transformation and Center-point Projection method to solve the auto-recognition of analogy measuring instruments.The instrument image is gained through CCD. Many image processing techniques such as Median Filter, Image Binarization and Image Hilditch Thinning algorithm, are applied to preprocessing image. Hough Transformation and Center-point Projection method are used to extract object information, ascertain the reading of analogy measuring instruments.According to the characteristic of automotive dashboard, an edge detection method using Hough Transformation and Center-point Projection method is successfully used in the feature extraction of a meter's image in image processing. A segmentation method combining grey level and structural feature in spatial domain is proposed to separate the objects from the noisy background accurately and fast. To obtain the center skeleton of binary valued image, a fast effective Hilditch thinning algorithm is proposed and gets good results in the location of the medial lines of scales and pointer in the meter's image, it improves the contrast between the background and object of image much more effectively. Experiments show that the proposed method works very well for the pointer detection. Comparing with the classical Hough Transformation, it has the advantage of fast processing speed and very low computation storage.A Computer Vision inspection system for automobile gauge panel is successfully developed on the basis of the research works above. It is used in the automobile assemble line, performing quality inspection task with high accuracy.