The detection of building wall surface defects is of great significance to eliminate potential safety hazards. In this paper, a research on building wall design defect image recognition based on partial differential equation is proposed. Collect the image data of building surface defects, sample and quantify the collected images, and preprocess the defect images such as digital threshold segmentation, filtering, and enhancement. Then, the improved partial differential equation is used to recognize the image as a whole. The second-order partial differential diffusion equation and the fourth-order partial differential equation are used to recognize the high-frequency and low-frequency bands of the image, respectively. The kernel principal component analysis algorithm is used to transfer the overall image input space to the high-dimensional feature space. The kernel function is used to calculate the inner product in different subband images of the high-dimensional feature space to reduce the dimension of the overall image. The processed coefficients are inversely transformed by nondownsampling contour wave to realize the overall image recognition and ensure that the edge information of the source image does not disappear. Experimental results show that compared with other algorithms, the proposed algorithm has better effect and better stability.
Silhouette Reconstruction of Sub-surface Defects Based on the Point-source Constrained Partial Differential Equation over the A0-mode Lamb Wave Field