Method for positioning paper money number image region

Abstract

The invention relates to a method for positioning a paper money number image region. The method comprises the following steps: acquiring a horizontally arranged single paper money image; acquiring the appearing region of the number of the current paper money image, respectively selecting two rows in each region, and acquiring the proportion of the quantity of pixels with the pixel values less than an upper threshold F accounting for the total quantity of the pixels; carrying out [-1, 0, 1] Sobel gradient calculation respectively on the pixels of each row in each region, and acquiring the proportion of the quantity of the pixels with the gradient calculated values larger than a lower threshold G accounting for the total quantity of the pixels; respectively calculating the sums of V1 and V2 of the sampling rows in the four regions, searching for the maximum value of the sums of V1 and V2, and taking the region corresponding to the maximum value as a number region; and amplifying the image of the number region, carrying out slant and longitudinal draw correction on the amplified image, removing the blank region of the image of the number region, and identifying characters of the image of the number region. The method disclosed by the invention has the advantages that the paper money number region can be accurately positioned and the reliability is high.

Description

technical field

[0001] The invention relates to an image positioning method. Background technique

[0002] The banknote image serial number area positioning method is a common step in serial number recognition, and currently published methods mainly include two types. The first type is the projection positioning method. The binarized image is projected horizontally and vertically, and the number of pixels in each row and column in the horizontal and vertical directions is counted to determine the number area. This method is affected by the noise points after binarization, and it is easy to locate upward or downward.

[0003] The second category is a positioning method based on rectangular frame search. Since the focal length and object distance are fixed when the sample image is taken, the width w and height H of the pure number area are also fixed. Create a rectangle with a size of w×H pixels, and count the total number of black pixels in the rectangular area with the cu...

Images