Gradient histogram angle rapid calculation method

Abstract

The invention adopts a simple and convenient mode to conveniently and rapidly calculate a gradient direction, and particularly provides a gradient histogram angle rapid calculation method, which comprises the following steps: S1, defining the direction of an image as an included angle A between gx and gy, so that the original gradient direction calculation mode is A = arctan (gy / gx), namely tanA = gy / gx, and gx and gy are gradients of the image in the x direction and the y direction at the (x, y) point respectively; s2, assuming that 0-180 degrees are divided into N directions, positive and negative are not distinguished, sign-free gradients are not distinguished, each n degree is one direction, [0-(n / 2) degrees) and ((180-n / 2)-180 degrees) are ori1, N is a positive integer larger than 1, the angle n of each direction is equal to 180 / N, and S3, due to the fact that the tan value of each boundary angle is known, the angle n of each direction is equal to 180 / N, and the angle n of each direction is equal to 180 / N. And directly judging whether the direction ori is the direction from the first direction to the Nth direction according to the value of tan A (dy / dx) based on the corresponding relationship.

Description

technical field [0001] The invention relates to the technical field of image processing, in particular to a method for quickly calculating gradient histogram angles. Background technique [0002] In the prior art, the usual method for calculating image gradients is: [0003] 1. Calculate the image gradient direction and magnitude [0004] For an image pixel (i,j): its gradient Hx(i,j) in the x direction and Hy(i,j) in the y direction [0005] Hx(i,j)=I(i+1,j)-I(i,j) [0006] Hy(i,j)=I(i,j)-I(i,j+1) [0007] The gradient operator sobel and Laplacian can be used to perform convolution calculations on the image. [0008] Thus, the gradient magnitude M(x,y) and the gradient direction θ(x,y) can be obtained: [0009] [0010] θ(x,y)=arctan(Hy / Hx) [0011] 2. Calculate the bin value [0012] The gradient direction θ is [0,180], which is divided into n bins. For example, n=9 in OpenCV means that a circle is divided into 9 parts, each part of the interval is for the bin v...

AI Technical Summary

Problems solved by technology

[0017] When calculating the HOG feature of the image, the gradient direction of the image will be involved. From the definition of the gradient direction of the image, arctan needs to be used to calculate the gradient direction value in actual programming, and this step is in a large loop, so the calculation is slow and complicated.

Images