PATTERN RECOGNITION TECHNIQUE APPLIED TO MINERAL RESOURCES EXPLORATION-- A CASE STUDY FROM BADSHAPUR AREA, JAIPUR DISTRICT RAJASTHAN

Abstract

1.1 The need of Pattern Recognition in mineral ex loration Since prehistoric time till now the importance of mineral is evergreen. Regardless of synthetic and substitute mat-rials, man's insatiable demand for basic resources will surely continue to expand., and. will accelerate as the populous underdeveloped nations begin to emerge. The types of raw materials that will be needed require careful scrutiny, but there is no doubt as to the need. As consumption grows, the incentive to find new resources to replace the mined material grows also. As new mines are put into production, favourable areas for economic resources become harder to find. It then becomes necessary to take greater risks on poorer indications, and to rely more heavily on technical methods for finding ore. No hypothesis or anomaly can ever rank as high as does visible ore in place, but such exploration guides are becoming increasingly more and more rare. Geophysics seldom gives a unique indication in the effort to discover Cc ncealed or obscured deposits of economic material. Moreover a single method of exploration after does not provide a unique answer. This is the reason why integrat-ed method of mineral exploration is carried out in a virgin 2 area. The need of a team comprising of geophysicist, geo-chemist and geologist is of utmost importance. Hence the development of a technique in mineral exploration utilising the services of geophysicist, geochemist and -geologist, at a time, for interpretation strategy will be a great boon to Earth Resources Technology. The purpose of this work is to design an efficient, fast and economic method of interpretation of multidiscipli-nary exploration data by Pattern Recognition Technique - a new born but a prospective baby in the world of technology. 1.2 What is Pattern Recognition In their widest sense, patterns are the means by which we interpret the world. What is the pattern recognition process then When a human glances at a printed page and recognizes character after character without hesitation, he is using fixed rules which he learned from experience. He has long since refined the discrimination of ' o' in a standard. type font to a fixed decision rule. He certainly could not explain that rule, but it clearly exists. He developed this ability from experience. At some point in time, he was repeatedly exposed to samples of the character 'a' and samples of the character 'o' and told which they were. From examination of these 'labeled' samples, he developed a decision rule. 3 There are two aspects to pattern recognition -(1) making an efficient space called feature space out of measurement space, (2) developing a decision rule, and using it. The actual recognition occurs in the use of the rule, the pattern is defined in thy, learning process by the labeled samples. In mathematical pattern recognition, we want a decision rule which can classify examples of patterns quickly. We may usually proceed with more leisure in learning the pattern, that is in deriving the decision rule. The pattern is defined by the labeled samples of that pattern. Samples are presented as examples of one class of pattern, (e.g. the letter ' ot) or another (e.g. the letter tat). The representation tot in the present context is a character of the alphabet, in another context it is a 'circle' rather than a 'square'. There is no way of discriminating the character from the geometric figure except by stating what we wish to decide, by defining the pattern classes. Mathematical pattern recognition provides a formal structure for solving problems of the sort described. When those problems can be posed within this structure and when an adequate number of labeled samples of the classes are available, the results can be dramatic. However, the techniques of mathematical pattern recognition are simply a collection of numerical algorithms for solving very particular problems posed in very particular ways. Any success in their applica-tion depends on careful formulation by the user and an 2 understanding of the assumptions involved in their use. Be-cause of the rather miraculous feats of sensory pattern recog-nition performed by humans, there is a tendency to expect auto-matic results from computer-based pattern recognition. As in any other practical problem, a great deal of thought in pre-paration of the data and in selection and implementation of methods is required, quick 'feasibility studies' in computer based pattern recognition usually produce quick inconclusive result.