Abstarct: For decreasing production costs and human failures, Iran industries need to automation and automatic control production process increasingly. With considering old measuring equipment’s in Iran industries and therefore big investment need to update them, automatic reading method for analogue measuring equipments baxsed on image processing is presented in this project. This measuring equipments divide to main types: first type is measuring equipments with pointer like pressure gauge and second type is semi-digital measuring equipment’s like gas counter. For reading analogue measuring equipment’s first, a photo from front view is taken and after that Wiener filter and unsharp mask is performed on the photo as a pre-process. The result is a soft Image with stand out edges. after that with applying Huff transformation circles in image will be recognized as monitoring screen and will be extracted from image and with use of canny method edges will be recognizes. With regarding to this monitoring screen and with use of a suitable limit of recognition, inappropriate edges will be removed. After that longest line in the image with use of Huff transformation will be recognized and will be considered as pointer. for reading semi-digital equipment’s with use of orthogonal observation, In this project gas counter will be measured. first Wiener filter and histogram stabilizer is performed on the photo as a pre-process. Then with use of horizontal histogram, the part of image that is related to counter will be extracted. after that with use of Vertical histogram, decimal part will be extracted. With examining intensity of light, six different level will be gained and baxsed on every level photo will be binary and all continuum components will be recognized and extracted. From every component, average block feature and histogram gradient feature will be extracted. This features will be learnt to a Artificial Neural Network with 11 information class learning system. Among these Information classes, 10 are assigned to numbers and one is assigned to invalid data. At last with mixing categories and numbers of founded values in every image, best binary image will be selected and value of analogue measuring equipment will be read.