A Study on Ancient Temple Structural Elements Recognition Using Genetic Algorithm

The systematic and scientific study of the lifestyle and culture of earlier peoples is known as archaeology. The Indian history of archaeology spans from the 19th century to the present status, it includes the region's history investigated by a wide variety of archaeologists. They don’t have any such authentic digital methods to be quoted in research. Manual Recognition of ancient temple structural elements is quite difficult to recognize, has archaeologists face many complex problems because they don’t have any automated Recognition method. Recognition is helpful for archaeologists to get more detailed information of ancient temples which is very important for further research. Thus, in this work it is proposed to develop automated method for Recognition ancient temple structural elements (vimana & pillars) for further archaeological research purpose. The proposed method extracts Genetic programming evolved spatial descriptor and classifies the temple structural elements visited by archaeologists based on linear discriminant method [LDA]. The proposed method is divided into 3 main phases: pre-processing, genetic programming evolution and Recognition. The Generalized Co-Occurrence Matrix is used in the pre-processing phase to change images into a format that genetic programming systems may use to process them. The second stage produces the best spatial descriptor to date in the form of a programme that is based on fitness Utilizing LDA, the Fitness is determined. Once the program's output has been received, it can be used for Recognition. Experimental results shows, it demonstrates relatively high accuracy in Recognizing both vimana(gopura) & pillars of different temples. The proposed method is implemented in MATLAB. These results will play very significant role in identification of temple architecture, which in-turn helps in conservation and reconstruction of temples. The proposed methodology will give 98.8% accuracy in pillars recognition and 98.4% accuracy in vimana recognition.