An Improved Encoder-Decoder CNN with Region-Based Filtering for Vibrant Colorization

Abstract

Colorization is the practice of adding appropriate chromatic values to monochrome photographs or videos. A real-valued luminance image can be mapped to a three-dimensional color image. However, it is a severely ill-defined problem and not has a single solution. In this paper, an encoder-decoder Convolutional Neural Network (CNN) model is used for colorizing gray images where the encoder is a Densely Connected Convolutional Network (DenseNet) and the decoder is a conventional CNN. The DenseNet extracts image features from gray images and the conventional CNN outputs a * b * color channels. Due to a large number of desaturated color components compared to saturated color components in the training images, the saturated color components have a strong tendency towards desaturated color components in the predicted a * b * channel. To solve the problems, we rebalance the predicted a * b * color channel by smoothing every subregion individually using the average filter. 2 stage k-means clustering technique is applied to divide the subregions. Then we apply Gamma transformation in the entire a * b * channel to saturate the image. We compare our proposed method with several existing methods. From the experimental results, we see that our proposed method has made some notable improvements over the existing methods and color representation of gray-scale images by our proposed method is more plausible to visualize. Additionally, our suggested approach beats other approaches in terms of Peak Signal-to-Noise Ratio (PSNR), Structural Similarity Index Measure (SSIM) and Histogram.

A Novel Unbiased Deep Learning Approach (DL-Net) in Feature Space for Converting Gray to Color Image

“…From visual perception, we can say that our method can produce more plausible images compared to others. These results show us that the proposed method trained with a Gray Deoldify [58] Iijuka [5] Larsson [46] Zhang [23] Zhang [24] Su [26] Gain [45] Prop. Meth.…”

“…al. [45], respectively. Besides, the average PSNR values of the proposed method are 7.81%, 3.50%, 6.50%, 7.15%, 5.04%, 7.77%, and 5.69%, which are higher than that of the methods in Deoldify [58], Iizuka et.…”

“…al. [45] proposed an encoder-decoder CNN model with filtering-based rebalancing techniques to colorize images. Larsson et.…”

“…From Figure 6, we see that the method in Deoldify [58] Gray Deoldify [58] Iijuka [5] Larsson [46] Zhang [23] Zhang [24] Su [26] Gain [45] Prop. Meth.…”

“…Zhang [23] forms desaturation in the second and over-saturation in the third images. Gain [45] form saturation in the seventh and eighth images. We find that there is no significant difference visually between the image generated from the proposed method and the ground truth.…”