Abstract: Solar flares occurring usually cause oscillations of nearby coronal loops. A precondition of studying these oscillations is to explore how to identify coronal loops. Meanwhile, accurate identification is also crucial to study and analyze the statistical features of the coronal loops and the relationship to coronal magnetic field. However, the traditional identification methods based on intensity and/or gradient threshold have some limitations, because coronal loops hold a relatively low contrast, blurred boundary, and the image noise is close to coronal loops. So we propose a new method to identify coronal loops using phase congruency and directional filter. The identification procedures are as follows: firstly, phase congruency technique is used to identify linear shape in a coronal image; secondly, a binary image is obtained from the image of phase congruency for detecting the loop features; thirdly, a directional filter based on Sobel Operator is applied to remove the redundant features that are nearly perpendicular to those target loops; fourthly, morphological filters are used to extract the shape of those loops; finally, quadratic fitting is used to smooth the identified loop features. For illustrating the performance of the method, the coronal loop oscillations taken with the Atmospheric Imaging Assembly on aboard the Solar Dynamic Observatory on September 11, 2011 are used to evaluate the identification and tracking process. The coronal loop oscillations were caused by a nearby X2.1 flare emission. We analyzed the identified results, and calculated their oscillation period, phase difference and decay coefficient. They are in good agreement with those of previous works. This implies that the proposed method can accurately identify the shape of a loop in an image; furthermore, the results can also be used to study the physical properties of coronal loop oscillations and the relationship to others solar phenomena.