The invention discloses an online detecting method for wizened defects of insides of walnuts. The online detecting method comprises the following steps: acquiring an image of a walnut under a dynamic condition by using an industrial CCD camera, carrying out binarization processing on the image to obtain a projection area of the walnut, and meanwhile acquiring weight information of the walnut under the dynamic condition by using a weighing sensor and a digital acceleration sensor; carrying out regression analysis on the projection area and the weight of the walnut to obtain a walnut weight prediction model, and calculating by using the walnut weight prediction model to obtain predicted weight of the walnut, and calculating relative error between the predicted weight of the walnut and the real weight of the walnut, judging whether the detected walnut is a wizened walnut or not by using the relative error as a judgment threshold value, and breaking the hull to verify judgment accuracy; carrying out regression analysis on different judgment threshold values and judgment accuracies corresponding to the judgment threshold values; and finally, searching a fitting function between the judgment threshold values and the judgment accuracies by using a golden section method, finding out the optimal judgment threshold value, if the relative error of the weight of the detected walnut is greater than the threshold value, determining that the walnut is the wizened walnut, and if the relative error of the weight of the detected walnut is smaller than or equal to the threshold value, determining that the walnut is an abnormal walnut. Nondestructive detection of wizened walnuts can be realized by the projection area of the images of the walnuts and the weight information of the walnuts only, and the method is simple. Compared with the prior art, the online detecting method has the advantages that detecting cost is greatly reduced, detecting speed and accuracy are improved, and the online detecting method is suitable for factory large-scale production, and can be used for online detection of wizened defects of the insides of various nuts such as the walnuts.