Matlab Vehicle / Car Tracking / Kalman Filter / Noise Reducing

I have the following challenge. I plotted a camera-based distance measurement of a vehicle over the number of frames. As you can see in the picture (y=distance, x=frames), noise causes measurement errors. The measured vehicle could theoretically accelerate or drive constantly. In this case it drives constantly.

Does anyone have an idea how I can reduce it to reduce this noise during a measurement? The noise distribution is almost constant corresponding to a distance. I am currently trying to find a solution with the Kalman filter in Matlab.

https://de.mathworks.com/help/vision/ref/vision.kalmanfilter.html

https://de.mathworks.com/help/vision/ref/configurekalmanfilter.html

The Kalman approach looks like this:

    ... % DistObj = Distance values from meausurement 

    if count < 6
        HoldD(count) = {DistObj};  
            count = count +1; 
    else
        HoldD(1) = []; 
        HoldD(end+1) = {DistObj}; 
    end 
    
    kalman = [];
    for idx = 1: length(HoldD)
       location = HoldD{idx};
       if isempty(kalman)
         if ~isempty(location)
           stateModel = [1 1;0 1];
           measurementModel = [1 0];
           kalman = vision.KalmanFilter(stateModel,measurementModel,'ProcessNoise',2,'MeasurementNoise',16); %1e-4
           kalman.State = [location, 0];
         end
       else
         trackedLocation = predict(kalman);
         if ~isempty(location)
           d = distance(kalman,location);
           trackedLocation = correct(kalman,location);
         else
           print('Missing detection');
         end

        plot(fr,trackedLocation,'co');
        hold on 
    %     plot(fr,DistObj,'ko');
    %     hold on 
    
        xlim([-1 nFrames])
        ylim([0 70])
        
       end
     end

Could somebody help me please?

Answer 1

[Sulaymon Eshkabilov]

Hi,

In your problem, you may employ the moving average filter: movavg() to reduce the influence of the noises.

Good luck.