Image Recognition and Parameter Analysis of Concrete Vibration State Based on Support Vector Machine

Shuai Wang; Aixun Wang; Xudong Fu; Keyang Wu; Tong Lu

doi:10.3791/65731

Journal

/

Engineering

/

Image Recognition and Parameter Analysis of Concrete Vibration State Based on Support Vector Machine

JoVE Journal
Engineering

A subscription to JoVE is required to view this content. Sign in or start your free trial.

JoVE Journal Engineering

Image Recognition and Parameter Analysis of Concrete Vibration State Based on Support Vector Machine

865 Views

•

08:27 min

•

January 05, 2024

DOI:

10.3791/65731-v

DOI:

10.3791/65731-v

•

08:27 min

January 05, 2024

•

15 Views

Shuai Wang^1,2, Aixun Wang², Xudong Fu¹, Keyang Wu², Tong Lu²

¹Wuhan University, ²Wuhan Construction Engineering Group Co. Ltd.

Transcript

The protocol described in this paper utilized the directional gradient histogram technique to extract the characteristics of a concrete image samples under various vibration cells. It employs a support vector machine for machine learning, resulting in imaging recognition method with minimal trained sample requirements and low computer performance demands. This approach significantly reduce the number of samples required, and lowers the computer performance requirements.

With a laptop equivalent at 2.3 gigahertz central processing units, the recognition process completes the train space differentiation of the supported vector machine within just 50 seconds. The image segmentation below size of 128 projects and 128 projects is utilized. The number of directional vectors for statistical angle inverse is set to 12.

In the 224 resolution image process, the best recognition occurrence for machine learning result is achieved.

Summary

The protocol described in this paper utilizes the directional gradient histogram technique to extract the characteristics of concrete image samples under various vibration states. It employs a support vector machine for machine learning, resulting in an image recognition method with minimal training sample requirements and low computer performance demands.