Extended Abstract
Introduction
Remote sensing science is one of the most powerful tools for the mineral explorations and mineral resource estimation. With regard to this science, any type of rocks with structural characteristics and mineral constituents has a special spectral signature, thus, using remote sensing techniques, different types of rocks in a particular area can be recognizable based on their reflective characteristics. Remote sensing techniques are considered as one of the standard methods in geological studies due to the identification of spatial patterns of rocks as well as their speed and economic price. Pervious geological studies indicate that the study area mostly contains basalt, limestone and marble, which has resulted in physical and chemical degradation of basalt stones under the influence of some geological events. Some parts containing basalt have lost their qualities due to these degradations. Therefore, the classification and separation of high-quality basalt zones from low-quality zones is the main objective of this paper.
 
Materials and method
The main objective of this study is to identify high-quality basalt zones in the Dir-o-Morreh mine located 50 kilometers from Tehran city near the lake of Hoz-e-Soltan. Basalt is a dark-colored and fine-grained igneous rock composed mainly of plagioclase and pyroxene minerals. Typically, this type of rock is formed externally or in the presence of air, such as the flow of lava, and these rocks can also take form intrusively like igneous dikes or narrow pillars. The basalt in the Dir-o-morreh mine is of igneous dike basalt type. In this study, the ASTER satellite multi-spectral images were used. These images allow us to have a good spatial and spectral resolution with regard to the objectives. However, reflectance conversion and atmospheric corrections were carried out on these images before using them, in order to enhance the accuracy of the project. Aerosols contained in the atmosphere are liquid or solid particles suspended in the air, which are very important in the evaluation of satellite imagery for remote sensing. After applying pre-processing, Basalt Exploration Index (BEI) was introduced and used to identify the basalt. The BEI index has been extracted using various sources, including the basalt spectral signature provided by the department of applied mathematics and statistics of Johns Hopkins University, ASTER satellite behavior (defined by the space team of NASA and Japan) and the Earth’s data which were collected to validate the results.  This index has been able to identify different basalt zones, including major extraction zones and other potentially possible zones. Moreover, this index is able to completely separate the basalt zones from the surrounding areas (mainly limestone, marble and clay rocks).  At the next step, convolution and morphology filters have been applied to separate high-quality Basalt zones from the low-quality. The amount of the brightness of an output pixel from the Convolution filters is a function of weighted average of the brightness of its surrounding pixels.  Using convolution with the selected kernel in satellite imagery returns a new filtered spatial image. High-pass Standard convolution filter was used in this study, which eliminates low frequencies of an image by retaining the high frequencies. The morphological nuclei used in this study are only the structural elements of this project and should not be confused with convolution kernels. In order to control the obtained results, the classified zones were double-checked on the field.
 
Results
The results obtained from the field studies and the identified zones are appropriately consistent with each other using the proposed index. Supervised classification was applied to improve the level of assurance and accuracy. Supervised classification is based on the idea that the user can select sample pixels in an image representing certain classes and then use image processing software using these educational samples as the referral for the classification of all other pixels in the image. This classification algorithm can be very effective and accurate and classifies satellite images in pixel-based or object-oriented form. Supervised classification can result in the preparation of two maps in two different classifications, which is has been done by using the Maximum Likelihood Algorithm. MaxVer or Maximum Likelihood is a statistical classification method that takes the weight of average value of the distance between the classes into consideration, using statistical parameters. To achieve sufficient accuracy, this algorithm requires a number of educational samples or pixels (more than 30). The primary classification includes 5 types of rocks or classes: high-quality basalt, low-quality basalt, limestone, marble stone, and clay which are designated on the map. In order to increase accuracy of the proposed method, the second map was prepared with 3 different classes (low-quality basalt, High-quality basalt, and surrounding rocks) in the second stage.
 
Conclusion
These maps help us in preparing a new BEI which is more accurate and more capable. It was also able to prove its capability in the latest ground operations and determining the most zones with high-quality basalt.