Search

This study aims to develop advanced and training-free full-reference image quality assessment (FR-IQA) models based on deep neural networks. Specifically, we investigate measures that allow us to perceptually compare deep network features and reveal their underlying factors. We find that distribution measures enjoy advanced perceptual awareness and test the Wasserstein distance (WSD), Jensen-Shannon divergence (JSD), and symmetric Kullback-Leibler divergence (SKLD) measures when comparing deep features acquired from various pretrained deep networks, including the Visual Geometry Group (VGG) network, SqueezeNet, MobileNet, and EfficientNet. The proposed FR-IQA models exhibit superior alignment with subjective human evaluations across diverse image quality assessment (IQA) datasets without training, demonstrating the advanced perceptual relevance of distribution measures when comparing deep network features. Additionally, we explore the applicability of deep distribution measures in image super-resolution enhancement tasks, highlighting their potential for guiding perceptual enhancements. The code is available on website. (https://github.com/Buka-Xing/Deep-network-based-distribution-measures-for-full-reference-image-quality-assessment).

In this paper, a joint decision tree and visual feature optimization rate control scheme for ultrahigh-definition (UHD) versatile video coding (VVC) is proposed. First, we design a new rate-distortion (R-D) model for UHD videos, and we establish a decision-tree-based multiclass classification scheme to improve the prediction accuracy of the R-D model by fully considering visual features. Second, based on the proposed R-D model, the globally optimal solution is obtained through convex optimization. Finally, we embed our algorithm into the latest VVC reference software, VTM 10.2. According to our experimental results, compared with the latest algorithm in VTM 10.2 and other state-of-the-art algorithms, our method can achieve significant bit rate reductions while maintaining a given peak signal-to-noise ratio (PSNR) or structural similarity index measure (SSIM).

Italian ryegrass ( Lolium multiflorum Lam.) is one of the most important forage crops in southwestern China. In 2018, a leaf spot was observed in a field of Italian ryegrass in Mengyang, Sichuan province, China (30.96925°N, 104.10223°E). From January to early March, this leaf spot developed sporadically and appeared as brown to dark brown lesions. In late May, this disease reached a peak with incidence up to 80% and appeared as reddish-brown necrotic spots with a grayish white to brown center. To isolate the pathogen, sections (0.5 × 1 cm 2 ) of 30 diseased leaves collected from 10 plants were surface-disinfested in 70% ethanol solution for 30 s, 5% NaOCl solution for 5 min, rinsed thrice in sterilized distilled water, air dried, plated on potato dextrose agar (PDA), and incubated at 25°C in the dark for 4 days. To obtain pure isolates, the single-spore isolation technique (Cai et al. 2009) was used. The conidial suspensions were diluted to a reasonable concentration, spread onto PDA, and incubated at 25°C in dark for 24 to 48 h, and then single germinated conidia were transferred onto new PDA plates (Cai et al. 2009). Nine pure isolates showing similar morphology were obtained for further study. Colonies on PDA were dark gray in the center surrounded by white to gray, with gossypine mycelia on the upper side, and red to dark red on the reverse side. Conidia were obclavate or pyriform, olivaceous to dark brown, with 0 to 6 transverse septa and 0 to 4 longitudinal septa, 13.2 to 55.0 (27.9) × 6.3 to 12.5 (9.8) µm. Conidiophores were septate, hyaline to olivaceous brown, either branched or unbranched, geniculate at the tip, 2.5 to 5.9 μm wide and up to 70 μm long. These morphological and cultural characteristics were consistent with the descriptions of Alternaria alternata (Fr.) Keissl. isolated from Apple (Elfar et al. 2018). To confirm the pathogenicity on Italian ryegrass, healthy plants (8-week-old) of cultivar Splendor grown in five pots filled with potting soil were spray-inoculated with conidial suspension (1 × 10 6 conidia/ml). Plants in another, five pots were sprayed with sterilized distilled water as controls. All pots were individually covered with transparent polyethylene bags for 5 days to maintain high relative humidity and placed in a greenhouse at 18 to 25°C. At 14 days post inoculation, symptoms typical of brown to dark brown leaf spots developed on the plants inoculated with conidial suspension, whereas no symptoms on the control plants. The pathogenicity tests were carried out three times. The same pathogen was consistently re-isolated from inoculated leaves and confirmed by morphological characterization as described above. To further identify this pathogen, isolate HMCH-9 (=CGMCC 3.19924) was selected as a representative for molecular characterization. Following Woudenberg et al. (2015), the internal transcribed spacer regions 1 and 2 and intervening 5.8S rDNA (ITS), glyceraldehyde-3-phosphate dehydrogenase ( GPD ), translation elongation factor 1-alpha ( TEF ), RNA polymerase second largest subunit ( RPB2 ), and Alternaria major allergen ( Alt ) genes were partially amplified and sequenced. Sequences were deposited in GenBank (accession nos. MH567106 for ITS, MH567107 for GPD , MH567109 for TEF , MH567110 for RPB2 , and MH567108 for Alt ). BLAST analysis of all these five segments showed >99.8% identity with those sequences of ex-type isolate CBS 916.96 of A. alternata (Woudenberg et al. 2015). To our knowledge, this is the first report of A. alternata causing leaf spot on Italian ryegrass in China. The accurate identification of this pathogen would be useful for the prevention and control of leaf spot on Italian ryegrass in the future.