Quantitative detection method for defects of wood knots
A quantitative detection method and quantitative detection technology, applied in neural learning methods, image data processing, image enhancement, etc., can solve the problems of low false detection rate and missed detection rate, low detection accuracy rate, high false detection rate, etc. False detection rate and missed detection rate, short training time, and good robustness
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
Method used
Image
Examples
Embodiment 1
[0067] A kind of timber knot defect quantitative detection method of the present embodiment, refer to figure 1 , including the following steps:
[0068] Step 1, establishing a quantitative detection model for wood knot defects;
[0069] Step 2, input the image data to be detected in the above-mentioned timber knot defect quantitative detection model;
[0070] Step 3, using the above-mentioned quantitative detection model of wood knot defect to detect the image data to be detected, and obtain the detection result.
[0071] The above-mentioned establishment of a quantitative detection model for wood knot defects specifically includes the following steps:
[0072] 11) Obtain an improved Faster R-CNN model;
[0073] 12) Collect training sample data;
[0074] 13) Adopt the improved Faster R-CNN model, and use the training sample data to train the improved Faster R-CNN model, so as to obtain the quantitative detection model of wood knot defects.
Embodiment 2
[0076] A kind of timber knot defect quantitative detection method of this embodiment, based on embodiment one, refer to figure 1 , using the improved Faster R-CNN model, using the marked training sample data to train the improved Faster R-CNN model, so as to obtain a quantitative detection model for wood knot defects, and obtaining the improved Faster R-CNN model specifically includes the following steps :
[0077] Determine the backbone network structure of the Faster R-CNN model. After many experiments, the backbone network structure of the selected Faster R-CNN is Figure 4 The network structure shown is the network structure proposed by the embodiment of the present invention. The structure of the first layer of the FasterR-CNN backbone network structure is "convolution + activation + pooling", and the convolution kernel size of the convolution module is is 3*3*32, the activation function of the activation module is ReLU, and its mathematical expression is: f(x)=max(0,x);...
Embodiment 3
[0080] A kind of timber knot defect quantitative detection method of the present embodiment, based on embodiment two, refer to figure 1 , collect training sample data, the training sample data is image data containing knot defects and image data without knot defects; mark the position of knots in each image in the training sample data; use the improved Faster R -CNN model, using the above-mentioned marked training sample data to train the improved Faster R-CNN model to obtain a quantitative detection model for wood knot defects.
[0081]Collecting the training sample data in this embodiment specifically includes the following steps. The training sample data includes image data containing knot defects and image data not containing knot defects, image data containing knot defects and image data not containing knot defects Obtained by the following steps.
[0082] Image data containing knot defects: use an industrial line array camera to shoot the surface of sawn timber, which i...
PUM
Login to View More Abstract
Description
Claims
Application Information
Login to View More