Salient Region Detection Improved by Principle Component Analysis and Boundary Information

Abstract

Salient region detection is useful for several image-processing applications, such as adaptive compression, object recognition, image retrieval, filter design, and image retargeting. A novel method to determine the salient regions of images is proposed in this paper. The L₀ smoothing filter and principle component analysis (PCA) play important roles in our framework. The L₀ filter is extremely helpful in characterizing fundamental image constituents, i.e., salient edges, and can simultaneously diminish insignificant details, thus producing more accurate boundary information for background merging and boundary scoring. PCA can reduce computational complexity as well as attenuate noise and translation errors. A local-global contrast is then used to calculate the distinction. Finally, image segmentation is used to achieve full-resolution saliency maps. The proposed method is compared with other state-of-the-art saliency detection methods and shown to yield higher precision-recall rates and F-measures.

Adaptive Parameterized AdaBoost Algorithm with application in EEG Motor Imagery Classification

“…These data is labelled as non-reduced features (500 x 200). Principal Component Analysis (PCA) [43] is used then for the reduction of each feature to around one-tenth of their original size (except the average power). So, the total size of the reduced feature vectors becomes 91 x200.…”

Moving visual focus in salient object segmentation

“…With the proposed method, three sections of experiments to show the position of the proposed method to the existing work. The existing algorithms that we employ to construct the baseline are (RC(region contrast) and HC(Histogram contrast) [15], FT [25], CA(Context-aware) [24], LC(Spatiotemporal cues) [23], GB(Graph-based visual saliency) [32], AC(low-level features of luminance and color) [12], SR [22], HFT(Hypercomplex Fourier transform) [33], IT(Laurent Itti proposed in 1998) [16], MVS(L zero smoothing and principle component analysis) [34]).…”

“…It includes 1000 nature images with ground truth those are labeled manually. To better illustrate the performance of RC [15], MVS [34], HFT [33], CRC , and MRC, we choose some images from Achanta's dataset and construct two different datasets. The images in one dataset, whose salient objects are in the center of images, and in another dataset, whose salient objects are not in the center of images.…”

Salient Region Detection Based on Spatial Weight Map

“…But this saliency map shows poor performance in noisy background. ChienChi Chen, Jian-Jiun Ding [13] considered both local and global features to highlight the salient objects. The limitation of this work is that, the saliency detection is only applicable at the center of the image.…”

“…The saliency map is detected and are correctly localized. The proposed framework is compared with five state-of-the-art saliency detection methods: SR [4], AC [5], FT [6], CA [7], PH [13]. In phase of L 0 smoothing filter we set smoothing parameter as 0.01 and parameter that controls the rate is 2.…”