Research And Application On Automatic Calibration Technology Of Analog Instrumentation

Pointer instrument widely used in real life, Existing Analog Meter Verification is basically done by manual. This will not only bring on labor wasted, but also brings visual fatigue. Pointer instrument based on machine vision technology research test only appeared in recent years, And because of the complexity of the environment and the diversity of the instrument, few examples can be put into use. Identification of existing test present a certain of environmental limitations,having a bad Anti-interference ability.Taking this as a start point. In this paper, making a exploration of automatic verification pointer instrument.After viewing a number of domestic and foreign papers. About existing poor Anti-interference capability of pointer instrument identification system verification. Made a number of improvements and new solutions. The main work is as follows:1. Instrumentation image preprocessing. According to the characteristics Instrument image "Ming background, dark goal". To strengthen the tick marks and pointer information, Separated from the background, Information must be extended in the low gradation region. In this paper, proposed piecewise linear transformation of basing on gray expectations to achieve strengthen of Tick marks and pointers, prominent target area. Also using wavelet analysis method to remove image noise meter, this method can remove noise while better preserve edges and image details.2. Instrumentation image segmentation.Improved the Otsu method, The threshold of Otsu method is global threshold, is part of a global thresholding method.The method is poor in adapting illumination nonuniformity. In view of this, In this paper, Otsu method is improved.After improving the method,the method has a better adaptability to illumination inhomogeneity.3. Getting pointer rotation center. In this paper, single image extraction pointer rotation center, Proposed a method of connected domain area threshold to get connected domain tick, calculate the scale line centroid. Using the least squares method to fitting tick centroid, fitting center of the circle as the center of rotation of the pointer.4. Meter reading research.In this paper,using distance discrimination method to achieve to get meter readings The method for instrument of Ticks are nonlinear distributed is very effective.