Application Of Image Processing In Meter Identification And Fabric Defect Detection

With the development of the industry in automatization and intelligence, the computer vision and image processing are widely used in industry,which are promoting the meter identification and the fabric defect detection to become two industrial applications. Different identification algorithms are proposed for two kinds of meters: the value-display meter and the indicator meter. A real-time identifying system for the indicator meters are developed based on the algorithms in this paper. Meanwhile, the key techniques of fabric defect detection are discussed.In the research on meter identification, the features of the digital on value-display meter are extracted using the region statistics algorithm and the nearest distance measure is used for identification. For the indicator meter, an algorithm based on the calculus operator for the identifying system of indicator measuring instrument is presented to overcome the deficiency of existing methods. First the region of arched scales is ideally located by the calculus operator. Then the arched scales are standardized into parallel scales and located by the image projection. It can be reach the demand of the real-time monitoring system and precision adjustment for its few operand and high precision.In the study on the fabric defect detection, the Gabor filters which have special features on orientation and scale are used to analyze the fabric images and the defects are located by comparing the output energies between the detected image and the normal image. For detecting the defects varying widely both in orientations and scales, the multi-channel Gabor filters whose discrimination is above 1 are selected by training, and all the filters responses data are fused as the final output. When the detected fabric defects have the steady orientations and scales, the output can be achieved only with the optimal Gabor filter that provides the best discrimination.