Desktop version

Home arrow Engineering arrow Cellular Image Classification

Discussion

To provide a more comprehensive analysis of the proposed LLDC method, we further evaluate its performance with respect to the number of nearest neighbors for calculating local distance vector and the coding process respectively. It should be noted that, the classification performance evaluated in this section is classification accuracy at the cell level.

Neighbor Number kLDV on Calculating Local Distance Vector: in our proposed method, we firstly introduce a merged manifold M for all the classes. Secondly, we transform the original local features to local distance vectors by searching the nearest kLDV neighbors around the local feature without regard to classes isolated from the local feature. Figures 5.3 and 5.4 show classification accuracy under various values of kLDV for ICPR2012 dataset and ICIP2013 training dataset respectively. Obviously, the proposed LLDC method achieves the best classification performance when kLDV = 35 while using LSC coding scheme for ICPR2012 dataset. For ICIP2012 training dataset, kLDV = 50 is the best choice while using LLC coding scheme.

Neighbor Number kLLC on LLC Method: we investigate the effect on classification performance under various neighbor number, kLLC, in approximated LLC coding scheme. Figure5.5 shows the performance using kLLC e {2, 5,10, 20, 30,..., 70}. As can be seen, the best classification accuracy is achieved when kLLC = 5 and kLLC = 60 for ICPR2012 dataset and ICIP2013 training dataset respectively.

Classification accuracy of the LLDC method under kiDV 6 {2, 5, 10, 15, 20, ..., 40} on ICPR201 dataset

Fig. 5.3 Classification accuracy of the LLDC method under kiDV 6 {2, 5, 10, 15, 20, ..., 40} on ICPR201 dataset

Classification accuracy of the LLDC method under kiDV 6 {2, 5, 10, 20, 30, ..., 70} on ICIP2013 training dataset

Fig. 5.4 Classification accuracy of the LLDC method under kiDV 6 {2, 5, 10, 20, 30, ..., 70} on ICIP2013 training dataset

Classification accuracy of the LLDC method using LLC strategy under kiic 6 {2, 5, 10, 20, 30,..., 70}

Fig. 5.5 Classification accuracy of the LLDC method using LLC strategy under kiic 6 {2, 5, 10, 20, 30,..., 70}

Classification accuracy of the LLDC method using LSC strategy under kisc 6 {2, 5, 10, 15, 20,..., 40}

Fig. 5.6 Classification accuracy of the LLDC method using LSC strategy under kisc 6 {2, 5, 10, 15, 20,..., 40}

Neighbor Number kLSC on LSC Method: with respect to LSC coding strategy, only kLSC nearest neighbors of a local feature are considered in coding procedure. We discuss the impact of different kLSC for staining pattern classification performance. Figure5.6 shows the classification accuracy under kLSC e {2, 5, 10, 15, 20,40}. Obviously, kLSC = 10 is the best choice for ICPR2012 dataset while kLSC = 30 is the best for ICIP2013 training dataset.

 
Source
< Prev   CONTENTS   Source   Next >

Related topics