OpenCV-based picture watermark automatic identification processing method and system

Abstract

The invention belongs to the technical field of computer vision, and particularly relates to an openCV-based picture watermark automatic identification processing method and system, which comprises an identification processing method, and the core steps of the identification processing method are as follows: establishing a picture shadow operation model, requiring the same picture resolution and the same format, wherein two pictures with the same watermark serve as input parameters, and the output parameter is a matrix with the same resolution as the input parameter; using algorithm preconditions: firstly, subjecting the background color to transparency processing, then subjecting images to graying processing; using a core algorithm, subjecting two image matrixes to subtraction to obtain an absolute value, then subtracting the absolute value from 255 to serve as a calculation result, and not calculating dark color pixels. According to the invention, extraction is directly carried out in the to-be-processed picture through an algorithm, and picture classification can also be automatically processed through a crawler program; in the algorithm, a grayscale image is used for watermark positioning, and an rgb value is used for picture watermark extraction, so that the accuracy of watermark picture pixel extraction is ensured to the greatest extent.

Description

technical field [0001] The invention relates to the technical field of computer vision, in particular to an openCV-based image watermark automatic recognition processing method and system. Background technique [0002] With the continuous development and growth of the company's e-commerce business, the platform needs to connect a large amount of product information for the first time when connecting with suppliers in ERP, especially product picture information. Suppliers often upload pictures of products sold on other platforms, causing the company Product compliance issues; In addition, the company's basic database construction also needs to check whether the pictures have watermarks in batches, and needs to remove the watermarks in batches. The company also tries to manually repair the watermarked pictures, but the workload is huge. [0003] After retrieval, the Chinese Patent Publication No. is CN108109124A, which discloses a method for repairing image watermarks in indet...

AI Technical Summary

Problems solved by technology

[0004] The above patents also have the following disadvantages: manual preprocessing is required for a large number of images with watermarks, watermark templates need to be manually marked and extracted for model training, and the manual processing workload is huge; grayscale and binarization processing is used in the algorithm, which will Loss of template pixel quality results in unclean cleaning of the watermark, with imprints; in normal watermark removal scenarios, it is difficult to obtain the original image without the watermark, especially for the need to remove the watermark from a large number of images, it is impossible to obtain the original image

Images