Study on the automatic instrumental reading system based on image processing

A new solution to the automatic reading of the traditional ocular optical instruments is proposed in this paper. A CMOS image sensor is used as the image capturing device and a series of image pre-processing algorithms are adopted to locate the reticles and digit characters in the ocular lens view images. A two-layer character recognition model which combines the template matching method and an optimized concurrent neural network is established. The method based on multiple complementary extraction is used to generate eigenvectors of the network. In order to improve the fault-tolerance capacity, rotation invariant features based on Tchebichef moments are extracted from digit characters and a four-dimensional group of the outline features is also obtained. The experimental result shows in applying the newly developed algorithm, the system has high recognition ratio and acceptable reading precision. In addition, the method has good universality and flexibility, which can be widely applied for the upgrading of various traditional optical instruments.