An experimental study to evaluate shapes and sizes of defect under thick coating by microwaves NDT is demonstrated. Specially fabricated thick fire protect coated steel panels with embedded defects are inspected using an X-band (8.2–12.4 GHz) open-ended rectangular waveguide. The fundamental idea behind using this probe is presented along with several experimental results to validate this method for defect detection under coating. The reflected signal related to the phase and magnitude of the reflection coefficient at the waveguide aperture is used to create images of these coated samples under test. These images indicate the ability of microwaves for identifying and sizing defects under thick coating layer. Linear sweep technique is used here to obtain multiple frequency spectrum variances. Principle Component Analysis (PCA) algorithms have been employed to enhance the resolution of our proposed method. A series of performance comparison with PCA algorithms are also provided to extract the defect features from thick coating layer influence. To evaluate the proposed technique, steel with known defect and five coated steel plates with unknown defect under different coating thickness are measured. Results indicate that the defect detection capability has been enhanced with the suitable use of signal processing methods.