Can a math filtering algorithm handle three coordinate points?

Filtering algorithms can handle three coordinate points. Application of filtering in three coordinates:

1, the impact of roughness on the measurement: the measurement point is also zoomed in the figure to obtain a large number of points, the surface roughness is considered to be, the cause of "noise".

2, the probe's mechanical filter:

Selection of the probe diameter - the use of the probe to measure the workpiece will be due to the influence of the surface structure of the workpiece to produce mechanical filter.

Because the probe diameter is too large fine workpiece surface shape can not be captured, so it can be regarded as mechanical low-pass filtering.

3, three coordinates of the filter:

With the same parameters for low-pass filtering of the scan line.

As shown below, the difference in the depicted graphs is not obvious.

4,2 RC filtering: The original standardized filter for roundness measurements is no longer used, but has been replaced by modern filter calculations.

5, Gaussian filter: coordinate measurement technology in the standard filtering algorithm. This filtering method for the standard algorithm is widely used. He uses a Gaussian curve to weight the measurement points to get a new profile.

6, spline filtering: enhanced filtering method based on filtering equations (polynomial computation), spline filtering is more in line with the standard, but also better than Gaussian filtering but is not the standard filtering method.

Expanded:

Image filtering is a very important image processing techniques, now the big fire of the convolutional neural network is in fact also a kind of filtering, are using the convolution kernel to extract the feature pattern of the image. However, traditional filtering, the convolution kernel used is a fixed parameter, which is manually designed by a very experienced person, also known as manual features. The convolutional neural network's convolutional kernel parameters are initially unknown, and are better adapted to different tasks as they are learned from the data and the neural network backpropagation algorithm.

Adaptive median filter

The median filter is a commonly used nonlinear filter, whose basic principle is to select the median of the values of pixels in a neighborhood of the pixel to be processed to replace the pixel to be processed. The main function is to make the gray value of a pixel closer to the pixels in the surrounding domain, so as to eliminate some isolated noise points, so the median filter can eliminate the pretzel noise very well. Not only that, the median filter can effectively protect the boundary information of the image while eliminating the noise, and will not cause much blurring to the image (compared to the mean filter).

The effect of the median filter is affected by the size of the filter window, and there is a contradiction between the elimination of noise and the protection of the details of the image: a smaller filter window can protect the details of the image very well, but the filtering effect on the noise is not very good, because the actual noise can't take up only one pixel position; on the contrary, a larger size of the window has a better filtering effect on the noise, but it will cause some blurring of the image. The image will be blurred to some extent. In addition, according to the principle of median filter, if the number of noise points in the filter window is larger than the number of non-noise pixels in the whole window, the median filter can not filter out the noise well.