Preprocessing (Feature) Pattern extraction with Gabor filter

Hi guys, I am currently looking for an appropriate way to filter a pattern from a very noisy environment. After a short search I came across the Gabor filter and then slowly approached the optimum (most accurate extraction) with the help of a Gabor filter bank. Since I would like to examine the pattern afterwards with the help of a gradient analysis, this way seemed promising to me for the time being. Now I have the results and was initially also very satisfied.

However, when I looked at the different intensities of the original image, I noticed that the intensities is strongly distorted especially at the end (400 px - 600 px) (see pictures: first original image, second gabor filtered image). The transition at about ~ 280px - 300 px should be preserved if possible. The peaks behind it are irrelevant

Now my question to you, are there perhaps certain adjusting screws, which I can still adapt or is Gabor generally the wrong approach. Are there perhaps other possibilities to strengthen the energies and at the same time to smooth the course ?

To the short explanation concerning the axes:

The x-axe indicates a measurement series. In a measurement series there are 12 images, which are taken in a dynamic scene. From this, the average value is formed over the entire measurement series. The camera is located at a static point and the scene is dynamic.

The y-axis indicates the respective pixel row over which the average value is calculated.

The z-axis indicates the intensity of the respective channel.

Here is an example of one image from a total of 12 images of a measurement series:

RAW GABOR