This paper is concentrated on the lengthened objects detection algorithm. This algorithm could be included to the structure of the autonomous navigation system on the flat landscape. The lengthened objects are the objects of the underlying terrain, which could be presented as the stripe or the border between typical terrains with different reflection characteristics. The first part is devoted to the description of this algorithm. The base of algorithm is statistical comparison of signal histograms. The criterion of the comparison could be referred to the maximin group of the criteria. The histograms present the discrete form of the probabilities and their intersection could be interpreted as some kind of the probability of terrain coincidence. The histograms are different for typical terrain types. Then we made an excursion to the implementation of the algorithm. It is discussed the typical variant of the implementation. The attention in the paper also attracted to the problem of histograms approximation. Here we show the algorithm errors, which should be eliminated or at least decreased. Some recommendations on the errors decreasing are provided.