A non-iterative edge-preserving image enhancement method is proposed for classifying human dermic tumors. Dermic tumor images typically feature more background residuals, which provide corrupted data for classification. In this instance, we separate the image data into two layers: Base layer and detailed layer. The artefacts of detailed layers were removed by applying bilinear interpolation. Additional amplification is used to produce high-pitched, detailed layer edges. The images are augmented with the enhancement ratio 1.5 and 2. To sharpen the image, additional Gaussian filtering is performed on the components of the detailed layer. The modified detailed layer's and the base layer’s components are merged to get the enhanced image. For image denoise and edge preservation, the enhanced image is fed to the bilateral filter. Quantitative and qualitative techniques are employed to evaluate the image enhancement. For classification we use Support Vector Machine and Convolution Neural Networks. Accuracy of 97.29 and 97.75 is obtained for the classifiers SVM and CNN respectively.