Structural　assessment　of　buried　energy　pipelines　is　often　hindered　by　the　abundance　of　external　vibrations　resulting　in　nebulous　noises.　Effective　and　secure　nondestructive　approaches　need　to　be　devised　to　efficiently　reduce　noise　in　multidimensional　magnetic　anomaly　signals　collected　from　a　pipeline.　This　study　focuses　on　the　mechanism　by　which　a　measured　source　signal　can　be　broken　down　into　low-　and　high-frequency　constituents　known　as　intrinsic　mode　functions　(IMFs).　By　doing　so,　a　well-defined　set　of　instantaneous　frequencies　is　obtained　utilizing　improved　complete　ensemble　empirical　mode　decomposition　(ICEEMDAN)　algorithms.　These　IMFs　contain　useful　structural　evidence　across　multiple　scales　that　can　be　extracted　for　effective　identification　of　the　defect　location.　To　accomplish　this　objective,　first,　the　signal　gradients　are　calculated　using　dual-density　complex　wavelet　transform　to　diminish　the　influence　of　the　geomagnetic　field.　The　multiscale　variance　fusion　(MVF)　algorithm　is　then　adopted　to　quantize　the　fluctuations　occurring　in　each　individual　IMF.　The　output　signals　generated　by　computing　the　variances　provide　sufficient　information　about　the　location　and　severity　of　the　pipeline　defects.　Numerical　simulations　for　a　buried　pipeline　model　have　been　presented　to　validate　the　authenticity　of　the　proposed　technique.　Indoor　laboratory　implantation　on　a　pipeline　test　sample　with　prefabricated　defects　justifies　the　effectiveness　of　the　ICEEMDAN-MVF　model,　to　localize　hidden　structural　flaws　in　energy　pipelines　without　physical　contact　and　even　in　more　complex　environments　with　multiple　sources　of　magnetic　interference.　©　The　Author(s)　2024.