The normalized intensity factor based on statistical first-order moment of gray-scale image is defined in this paper. The intensity factor can be used to distinguish the brightness level of a gray-scale image and to determine a threshold value for image segmentation. According to the intensity factor and the characteristic of human body in the gray-scale infrared image, a new algorithm of calculating the intensity-level threshold is designed which can be used for segmenting human body area in an infrared image. In the algorithm, based on the concept of intensity factor, a histogram of low brightness gray-scale image (LGIRI) is divided into three parts: a low-intensity region (0.25[Formula: see text][Formula: see text]), a medium-intensity region (0.25–0.75[Formula: see text][Formula: see text]), and a high-intensity region (0.75–1[Formula: see text][Formula: see text]), and then the intensity [Formula: see text] which satisfies the [Formula: see text] is selected as an intensity-level value [Formula: see text], and the intensity [Formula: see text] which satisfies [Formula: see text] is selected as an intensity-level value [Formula: see text], at last [Formula: see text] is the pixel classification threshold (the intensity-level threshold). It is noted that there is no preprocessing for image noise filtering and/or processing, and all images come from OTCBVS. Compared with the method of selecting trough points of the histogram as the intensity-level threshold, this algorithm avoids the problem of nonexistence of evident trough point at the high-intensity level of a histogram. Also, the experimental results show that the segmenting results of LGIRI processed by the algorithm are better than those of Otsu method.

Sentiment classification is attracting more and more attention because of its great benefits to social and human life. Usually supervised classification approaches perform well in sentiment classification, but the performance decreases sharply when transferred from one domain to another domain. In this paper, we propose an approach, SentiRank, which integrates the sentiment orientations of the documents into the graph-ranking algorithm for cross-domain sentiment classification. We apply the graph-ranking algorithm using the accurate labels of old-domain documents as well as the “pseudo” labels of new-domain documents, and investigate their relative importance for cross-domain sentiment classification. The experiment results indicate that the proposed algorithm could improve the performance of cross-domain sentiment classification dramatically.