The results shown in Fig. 5.1 were obtained for the first type of noise (salt and pepper).

The presented results show that the individual feature measurement errors can vary within wide limits. The maximum values of errors are shown in Table 5.1.

The table shows that salt and pepper noise has the least impact on features w(8) and from w(10) to w(13), whereas the largest on features from w(5) to w(7) and w(9). These differences arise from the measurement idea outlined in the previous chapters and thus their varying sensitivity to noise. Attention should be also paid to the degree of noise in images in Fig. 5.1d for ns = 0.01, ns = 0.1 and ns = 0.48 and the correlation with the error values shown in the graph in Fig. 5.1a. The graph in Fig. 5.1a can be divided into two specific areas: an increase in the value of error δ _w for ns ∈ (0, 0.025] and its decrease for ns ∈ (0.025, 0.5). Seemingly, a higher noise value results in a smaller feature measurement error. However, only the feature values are taken into account and there is no interpretation of the accuracy of the data on whose basis they are calculated. For example, Fig. 5.2 shows changes in the contour shape of the image L _H that forms the basis for the calculation of the features w for ns ∈ {0, 0.001, 0.003, 0.005, 0.01, 0.02, 0.05, 0.1}.

Stay updated, free articles. Join our Telegram channel

Join