One of the most important problems to address when applying interferometric techniques to industrial applications is the prevalence of noise, which results in poor fringe patterns and thus poor measurements. One of the techniques that suffers most from this problem is conoscopic holography. Though this interferometric technique is ideal for industrial inspection, the poor quality of fringe patterns obtained in adverse environments may make measurement impossible. Classical filtering techniques based on one-dimensional filters or general speckle removal filters, such as the Frost or the gamma, may not suffice in adverse conditions. Therefore, a new approach, based on the nature of the fringe pattern information itself, is worth looking into. In this paper we propose the use of orientational filters to develop a filtering method that not only removes noise of any nature, but also enhances the fringe-pattern information. Several approaches to these algorithms are implemented and evaluated using synthetic conoscopic fringe patterns under different noise conditions, showing how they clearly outperform classical filters with negligible distortion even in the worst conditions. Examples with real data acquired with the latest prototype of a conoscopic long-standoff profilometer are also provided.