Miniature parts are applied more broadly as instruments and equipments are getting more miniature. Imaging measuring methods used in miniature parts have become an important technology which is developing competitively at home and abroad. The methods are able to solve the limitation of routine measurements. This paper is based on image measuring methods and the image measuring apparatus for micro parts. There are many algorithms which are employed to appraise definition nowadays; however

different algorithms may get different results in different measuring systems and there is not a well-rounded algorithm to tell user how to make choice. So it is not easy to select a certain algorithm according to the characteristic of the appointed apparatus. The aim of his paper is to present an algorithm to deal with the selection of definition and the process in this paper shows typical ways of orientation for Telecentric systems. The procession of the experiment can be mainly divided into three parts. Firstly

the Telecentric image measuring system is used to grab image at different working distances; and the distance between the camera and the benchmark is recorded by a laser interferometer with double frequency. Secondly

the correlative algorithms including Brenner function

Tenengrad function

Gradient square function and the four modifier formula

entropy function are employed to appraise definition in this paper. Finally

the principal component analysis and the factor analysis are employed to analyze the data so as to work out the algorithm which is suitable for the apparatus from the above 8 algorithms and to work out the overall merit algorithm for definition. As a result

the suitable definition algorithm is worked out to be Tenengrad function and the overall merit algorithm for definition is presented. The synthesis evaluating model of definition has also been built in this paper. The data mining algorithm submitted in this paper is able to make choice in different evaluations when aiming at the exclusive requirement and to decrease the amount of the operands which will be used in next programs. The innovation of this paper is that it presents a whole new algorithm to deal with the selection of definition algorithm and a whole new algorithm to overall merit the definition algorithms. The experiment shows that the image grabbing system which using Telecentric systems would capture the most clear picture form series of pictures by using the algorithms.