Fusing directional wavelet local binary pattern and moments for human action recognition

Abstract

Recently, transformation‐based methods have been widely used in many computer vision areas because of their powerful representation ability. One of the most widely used transforms is the wavelet transform that has proved to be very useful in many applications. In this study, a new method for human action representation and description is proposed. This method combines the advantages of local and global descriptions. The method works by fusing the Hu invariant moments as global descriptors with a new local descriptor that is based on three‐dimensional stationary wavelet transform and the concept of local binary patterns. The performance of the new method was examined in two different ways. The first one is by fusing the proposed directional global and local features in one feature vector, while the other is using the features of different directional bands separately to train multiple classifiers and then using a voting scheme to vote for the best match. The performance of the proposed method is verified using standard datasets, achieving high accuracy in comparison with state‐of‐the‐art methods. In addition, the proposed method is proved to be robust to the changes in lighting and scale variations, but it exhibits limitations towards dynamic backgrounds.

Robust Image Forgery Detection Using Point Feature Analysis

“…First, There were few f eatures extracted in the copy-move algorithm from some of the images in the dataset using our 2 feature extraction methods; SIFT and SURF. One proposed solution can be using another feature extraction as block-based methods such as DCT [24] or DWT [3]. Also, our algorithm depends on dividing the images into blocks in the copy-move detection, however some objects can be divided between multiple blocks which can cause negatively affects the matching point step that compares the features of the blocks to one another.…”

Advances in human action recognition: an updated survey

“…Besides, the discussed supervised methods above, unsupervised approach such as k ‐NN has also been applied in HAR. Some of the works that utilised it are Devanne et al [80], Belhadj and Mignotte [18] and Al‐Berry et al [41]. Vishwakarma and Singh [17] adopted a hybrid classifier in which SVM classier was cascaded with 1‐NN classifier in order to increase the performance.…”

“…In this approach, the texture‐based features are used to represent either global or local features and they are usually combined with local or global features, respectively. An example of this approach is the work done by Al‐Berry et al [41]. For the local features, they computed the LBP after transforming blocks of 3D spatiotemporal video sequence using the 3D stationary wavelet transform.…”

“…For cross‐view learning, they implemented k ‐NN for action recognition. Decision tree classier has also been reported with an outstanding performance [41]. Another class of classifier, which is based on sparse representation, known as the collaborative representation classifier has also been applied in HAR [11, 81, 82].…”

Dense SIFT–Flow based Architecture for Recognizing Hand Gestures

“…In human computer interaction systems, the present and widely used trend is dynamic texture based techniques [4,5] and multimodal feature based frameworks [6,7] because of its powerful representation. One of the most broadly used features is local binary pattern (LBP) introduced by Ojala [8] for texture classification, that proved to be useful in computer vision applications.…”