NDT (Nondestructive Testing) is one of the most important tools for quality assurance of structural member during manufacturing stage and operational life. Among number of NDE methods, Ultrasonic has been the most important tool of choice to inspect, since the likely defects and material properties are easily and inexpensively uncovered by it. Manual ultrasonic is much better than radiography for planar defects, though the result is operator-dependent. Due to hazardous nature of X-rays, ultrasonic NDT gained importance in Medical and Quality checking departments. Due to limitations in accessing other side of the structure/specimen and immeasurability of transmitted signal (such as cardiovascular imaging), ultrasonic reflection imaging has great importance for NDE as well as medical applications. In the present paper, the ultrasonic reflection imaging of defected portion is used to characterize a defect by the Time of Flight data of reflected signal. It involves the reconstruction of slowness field for centered square, circular defects of epoxy implanted in Mild steel specimen. Backwall Reflected signal time of flight data is collected by using pulsar receiver ultrasound instrument in pulse echo mode. Slowness field reconstruction is done by Gordon’s Algebraic Reconstruction Technique (ART). Before the reconstruction, the signals are denoised by adopting EMD technique. Variation of parameters like number of iteration required for convergence, convergence rate, and normalized error with relaxation parameter have been studied. From the point of view of computation time, convergence and quality of reconstructed image optimal selection of relaxation for best performance has been chosen. Results have shown that algorithm is capable of characterizing defects; but for curve boundary defects, shape accuracy is less.