為了將影像完整呈現於各種尺寸的輸出裝置，且盡量減少視覺上的 扭曲變形，許多基於內容之影像尺寸調整機制被提出，如何有效地評估 各種方法的效果成為一項重要任務。本研究提出一個基於物件重要程度 的影像尺寸調整評估機制，透過語義分割方法將影像中的所有像素點分 類，根據語義中的類別，給予該所在區域不同的視覺重要程度，依此做 為人眼視覺對於該區域受破壞的敏感度衡量，希冀獲致更貼近使用者主 觀感受的顯著圖，並將其應用於長寬比相似性畫質衡量演算法以提升準 確度。我們另外觀察到人眼觀看無前景物影像時容易受到畫面整體資訊 損失的影響，因此提出無明顯前景物資訊損失懲罰調整策略。我們先利 用語義資訊判斷場景中有無明顯前景物，再給予不同大小級別的資訊損 失懲罰，提高無明顯前景物場景的評分準確度。實驗結果顯示，本研究 能有效評估影像尺寸調整機制，與現有方法相較有更高的準確度。 ;Many image retargeting methods have been proposed to resize images to fit in various sizes of display devices with less perceptual distortion. Assessing the quality of retargeted images has thus become an important task for developing such methods. In this research, we propose an image retargeting quality assessment (IRQA) based on importance of objects. We utilize semantic segmentation to classify pixels, which are assigned with different importance values representing the sensitivity of human eyes to distortion. A visual saliency map is created to better fit the subjective perception of humans and is then used in the evaluation method called “Aspect Ratio Similarity” (ARS) to improve its accuracy. Furthermore, as observing that human eyes tend to be affected more by the global information loss in images in which there is no obvious foreground object, we propose the strategy of information loss adjustment in such images. We first utilize semantic information to determine whether a foreground object exists and then adopt different degrees of information loss penalty to improve the accuracy of the assessment. The experimental results show that the proposed approach is effective in evaluating the image retargeting methods and outperforms existing quality assessment methods.
