Color image watershed segmentation method based on power law distribution

Abstract

The present invention discloses a color image watershed segmentation method based on the power law distribution. The method comprises the following steps: Step S1: obtaining an original gradient imageof a color image by using a SOBLE operator, and performing morphological minimum calibration on the original gradient image, thereby obtaining the calibrated image; Step S2: performing watershed pre-segmentation on the calibrated image so as to obtain a pre-segmented tag image, and selecting an area threshold through the power law distribution relationship of the region area-number; and Step S3:introducing the selected area threshold to destruct the self-organized critical state of the pre-segmented tag image, and redistributing the region in the pre-segmented tag image to reach a new self-organized critical state, thereby completing the final segmentation. The method disclosed by the present invention realize the advantages of suppressing over-segmentation and reducing the time complexity of the algorithm.

Description

technical field [0001] The invention relates to the field of image processing, in particular to a color image watershed segmentation method based on power-law distribution. Background technique [0002] At present, the watershed algorithm is widely used because it can quickly obtain continuous closed object boundaries, but due to factors such as irregular disturbances and noises in the image, it will have a negative impact on the segmentation results, that is, over-segmentation. Therefore, effectively reducing the over-segmentation problem of the watershed algorithm has always been the focus of research in this field. There are three main methods to suppress over-segmentation in the watershed algorithm: the first is to process before image segmentation, such as marking methods based on various filters; the second is to process after image segmentation, such as based on adjacent regions Various regional merging methods for similarity; the third type is manual intervention fo...

AI Technical Summary

Problems solved by technology

Using the first type of method is very efficient, but it is difficult to achieve a good segmentation effect due to inaccurate positioning of light and blurred edges

The second type of method has a good segmentation effect, but due to the high time complexity of performing multiple cycles, the effect is still unsatisfactory

The third type of method requires human subjective intervention, and the subjective judgment of different people will affect the objective segmentation results, which is not universal

[0007] There are problems of over-segmentation and time complexity of the algorithm in image segmentation in the existing technology

Images