In this paper, we present a computational model for recognition of steel type i.e., ferrite-pearlite and martensite-austenite grade steel by image processing of scanning electron microscopy (SEM) micrographs followed by phase identification, segmentation and quantification. For this purpose a comprehensive SEM microstructure database were produced through in-house experiments. The diversity in the database in terms of volume fraction, grain refinement and resolution in the microstructure images were achieved through annealing, normalizing and quenching treatments to four different plain carbon steel, containing 0.1, 0.22, 0.35 and 0.48 wt. %C and capturing the images at different magnification (500×, 1000×, 1500×, 2000×, 3000× and 5000×) in SEM under secondary electron (SE) mode. The image processing schedule for this work was developed employing the Gabor filtering, local binary pattern (LBP), random decision forest (RDF) and Otsu thresholding techniques for the purpose of noise reduction, statistical feature extraction, classifier design and segmentation respectively. The prediction accuracy of steel type was found to be significantly high. In case of ferrite-pearlite type steel, the predicted pearlite fractions for the steel containing 0.1, 0.22, 0.35 and 0.48 wt. %C were found to be 0.23 ± 0.02, 0.3 ± 0.03, 0.55 ± 0.03 and 0.65 ± 0.03 respectively for the investigated range of magnification (500×–5000×). Similarly, in case of martensite-austenite type steel, the predicted fractions of martensite for the steel containing 0.1, 0.22, 0.35 and 0.48 wt. %C were found to be 0.59 ± 0.02, 0.52 ± 0.01, 0.45 ± 0.01 and 0.56 ± 0.02 respectively within the magnification range of 500× to 5000×.