Object depth data-based object recognition algorithm and device

Abstract

The invention relates to a three-dimensional object recognition algorithm, in particular, an object depth data-based object recognition algorithm and an object depth data-based object recognition device. The objective of the invention is to solve problems in the prior art. According to the object depth data-based object recognition algorithm, object three-dimensional depth data with different translation and zoom factors are utilized to realize fast recognition of an object. The algorithm includes the following steps that: a world coordinate system is established, equidistant parallel line scanning is carried out, and scanning lines are projected into the world coordinate system, the scanning lines are spliced through rigid translation in a projection plane, a template and the depth data of a target object are obtained, template feature lines I and target feature lines IO are established; feature line vector extraction is performed the template feature lines and the target feature lines, and the similarities of the sub-line segments of the template feature lines I and the target feature lines IO are calculated respectively; and sub-line segments of which the similarities are larger than a set threshold value theta are adopted as matching results, wherein the sub-line segments have largest similarities, otherwise, matching fails, and the target object is different from the template.

Description

technical field [0001] The invention relates to a three-dimensional object recognition algorithm, in particular to an object recognition algorithm and device based on object depth data. Background technique [0002] Object recognition technology is a basic research of computer vision, and it is also a core part, which has applications in industry, military, medical, and other fields. Google, for example, is already testing smart devices equipped with 3D object recognition. It obtains stereoscopic images of surrounding scenes through a depth sensor, and can display 3D stereoscopic reconstruction scenes on the device screen. Users can easily find the products they want to buy while looking at the device. In terms of commercial use, it is expected to be able to display the location of sorted goods in the distribution center, thereby improving work efficiency. Chipmaker Qualcomm is developing a stereoscopic image recognition chip for robot-mounted platforms, focusing on drone ...

AI Technical Summary

Problems solved by technology

The selected features remain invariant to rotation, scaling, and brightness changes, and maintain a certain degree of stability against viewing angle changes, affine transformations, and noise, but the algorithm sacrifices storage space and computing time in exchange for the uniqueness and matching of feature points. Accuracy, and some feature points extracted do not have intuitive visual meaning

[0006] In addition, the above matching algorithms are all based on 2D image data and cannot be directly applied to the matching of 3D depth data.

Images