A Visual Method of Measuring Railway-Track Weed Infestation Level
Round 1
Reviewer 1 Report
The manuscript proposes a method to analysis the weed infestation level with the use of a vision system. The algorithm is simple but feasible, which is demonstrated with experiments. However, as a research article, the description of the method and the system is more or less too concise. I’d like to see more methodological and technical details.
- What is the usage of the additional light source in Fig. 2? It seems that all the presented images are captured under sunlight.
- How do you get the standards of green, light green and dark green presented in the III, IV and V rows of Table 1? It looks like that they are not pure color but have specific color distribution within the 8*8 pixels.
Author Response
Comment:
What is the usage of the additional light source in Fig. 2? It seems that all the presented images are captured under sunlight.
Answer:
The main purpose in the research of the use of an additional light source is not to illuminate the test object, because it is recommended to perform measurements in day, and to reduce the effect of camera glare when leaving the tunnel or under the viaduct. The camera operating in the automatic selection of recording parameters while driving through a dark tunnel extends the exposure and increases the sensitivity of the matrix. This results in blinding when it comes into the daylight. Additional lighting reduces this effect, as the changes in image recording parameters while driving through a dark tunnel are smaller. The text of the article has been complement with important information.
Comment:
How do you get the standards of green, light green and dark green presented in the III, IV and V rows of Table 1? It looks like that they are not pure color but have specific color distribution within the 8*8 pixels.
Answer:
Authors used as a patterns fragments with dimensions of 8x8 pixels taken from several images. These is results cause the different shade of green. The text of the article has been complement with this information.
Reviewer 2 Report
1) The authors write: "An additional light source was mounted next to the camera to facilitate recording images when the measuring flatcar passed through lower visibility places, tunnels for instance"
but all the images shown are outdoors with good visibility, an example of low visibility image would be good
2) In figure 1 a too simple Block diagram of the algorithm is shown, a more detailed description of the algorithm is due.
Author Response
Comment:
The authors write: "An additional light source was mounted next to the camera to facilitate recording images when the measuring flatcar passed through lower visibility places, tunnels for instance" but all the images shown are outdoors with good visibility, an example of low visibility image would be good.
Answer:
Weed is rare in tunnels, the main purpose of using an additional light source is not to illuminate the test object, as the measurement should be performed in day, but to reduce the camera glare effect when leaving the tunnel or under the viaduct. The camera operating in the automatic selection of recording parameters while driving through a dark tunnel extends the exposure and increases the sensitivity of the matrix. This results in blinding when it comes into the daylight. Additional lighting reduces this effect, as the changes in image recording parameters while driving through a dark tunnel are smaller. The text of the article has been complement with important information.
Comment:
In figure 1 a too simple Block diagram of the algorithm is shown, a more detailed description of the algorithm is due.
Answer:
The description of the algorithm has been extended.
Reviewer 3 Report
The work could be interesting. More technical contents and comparison are required. Significant improvement is required as list below.
- Colour and texture feature based inspection need to be reviewed and discussed. More robustness methods should be reviewed e.g. handling illumination influnce of colour G Finlayson, etc., Illuminant and device invariant colour using histogram equalisation, Pattern recognition 38 (2), 179-190, 2005;
- Mind difference of Tables and Figures. more tecnical contents and comparison should be provided with quantitative analysis.
- Mind major contribution. The abstract, conclusion should be refined with highlighting of contribution and condition based maintenance e.g. vegitation and Rolling contact fatigures of railtracks.
Author Response
Comment:
Colour and texture feature based inspection need to be reviewed and discussed. More robustness methods should be reviewed e.g. handling illumination influnce of colour G Finlayson, etc., Illuminant and device invariant colour using histogram equalisation, Pattern recognition 38 (2), 179-190, 2005;
Answer:
The procedure for normalizing image parameters, suggested by the reviewer and presented in the article, basically concerns the situation of equalizes of the colour dominant coming from light sources with different colour temperatures. In this situation, the equalizes will lead to the normalization of the colours of the recorded image. The authors thinks that the method, proposed by the reviewer, there does not apply to the image analysis proposed in the article. First of all, the image was recorded in the mode of automatic adjustment of the camera to the lighting conditions, which to some extent normalizes and equalizes the parameters for individual frames. Secondly, sunlight changes the colour temperature within wide limits, but without additional measuring devices it is not possible to determine the current colour temperature of the light. If we have this information could be the basis for the normalization procedure in a situation when the recorded image has colour dominants derived from the colours characteristic of the track surface and objects of the railway track area. The authors tried to normalize the lighting level for individual RGB channels but this works conducted to the falsified colours of the recorded image, which made it impossible to perform a base image analysis.
Comment:
Mind difference of Tables and Figures. more tecnical contents and comparison should be provided with quantitative analysis.
Answer:
Tables and figures are formatted as required by the journal. In further work, the authors will be conduct a more detailed analysis with use into account the all RAL green palette.
Comment:
Mind major contribution. The abstract, conclusion should be refined with highlighting of contribution and condition based maintenance e.g. vegitation and Rolling contact fatigures of railtracks.
Answer:
The defects rails kind of RCF are not affected by track weed infestation. The research were conducted in a non-contact track as well as in a classic track. In the classic track, the gap between the rails works due to temperature changes, so in this gap the weed infestation practically does not occur in it. In the case of a supported joint, in particular on wooden sleepers, weed infestation may occur in the joint area, but in the area of sleepers.
