An enzyme numerical membrane system for an NCC image matching algorithm

Abstract

The invention discloses an enzyme numerical membrane system for an NCC image matching algorithm, the system comprising: a surface layer membrane, which is used for storing an image to be matched, a template image and an initialized enzyme, and simultaneously outputting a position coordinate for calculating a matching position of the template image in the image to be matched; a ComputeT film for calculating parameters required for NCC calculation of a template image; a ComputeMatchingCood film for calculating a matching position of a template image in an image to be matched; and several ComputeNccValuelk films, which are used to calculate in parallel the NCC value of the template image at each position in the original image after each position is moved. The membrane system of the inventionhas strong parallelism, the template image does not need to be sliding matched on the original image, but simultaneously calculates the NCC values with all the sub-images, so that multiple NCC computations consuming time under serial computation are executed in parallel, and the computation time complexity is constant, independent of the data scale, thereby greatly reducing the time complexity ofthe entire NCC image matching algorithm.

Description

technical field [0001] The invention belongs to the technical field of image matching, and in particular relates to an NCC image matching algorithm enzyme numerical membrane system. Background technique [0002] The NCC algorithm is a classic algorithm in image matching. The definition of image matching means that there is a target image T, and the region that is the same target as T is found in the original image. Assuming that the size of the original image I is M×N, and the size of the template T is m×n, the traditional NCC image matching principle is: starting from the coordinates of the original image (0,0), the template is on the original image from left to right, from top to Swipe down, and every time you slide to a pixel, calculate the ncc value between the template and the sub-image corresponding to the current image, store the ncc values ​​calculated at all positions, and take the position corresponding to the maximum value of ncc as the template T in the original ...

AI Technical Summary

Problems solved by technology

Assuming that the size of the original image I is M×N, and the size of the template T is m×n, the traditional NCC image matching principle is: starting from the coordinates of the original image (0,0), the template is on the original image from left to right, from top to Swipe down, and every time you slide to a pixel, calculate the ncc value between the template and the sub-image corresponding to the current image, store the ncc values ​​calculated at all positions, and take the position corresponding to the maximum value of ncc as the template T in the original image The matching position on the , such as figure 1 As shown, a total of (M-m)*(N-n) NCC needs to be calculated. It can be seen that under the existing serial computing framework, the time complexity of the NCC algorithm increases with the resolution, and the calculation time increases polynomially.

When the resolution of the image to be matched and the template image is high, the existing serial architecture is difficult to achieve real-time calculation

Images