Abstract: Technologies currently used cannot effectively detect dehiscence furrow in Chinese dates, so automatic judgement of their apparent quality is a difficult problem. In this paper, the method for detecting dehiscence furrow using wavelet transform-based multiscale edge detection and mathematical morphology was proposed. According to the method, local mutation in Chinese date images was detected well and dehiscence furrow in Chinese date can be detected by combining with multiscale information. First, Chinese date images were processed to find the local maxima of gray grads by using multiscale wavelet function; Second, high and low thresholds were derived by self-adapting method and local maxima was divided up to obtain corresponding edge; Last, the detection results were derived from connection and erosion operation of mathematic morphology. Computer simulation results of multiscale edge detection for some images of Chinese dates with cross dehiscence furrow were presented, and more continuous, complete and single-pixel-width images were obtained. The method is useful for detection and classification of fruits.