WAVELET TRANSFORM APPLICATION FOR IMAGE PROCESSING IN MICROCONTROLLER BASED INTERNET OF THINGS SYSTEMS

Abstract

Purpose. Investigating the efficiency of Haar Wavelet Transform as an additional transformation to JPEG image compression in the context of limited resources on microcontrollers.

Methodology. The aims are exploring the effectiveness of using Haar wavelet transform as an additional transformation to JPEG for image compression, considering the limited resources of microcontrollers. The research focuses on reducing of the image transmission time and energy consumption by utilizing the combined approach. The methodology involves experimental evaluation of image quality processed using Haar wavelet transform and compressing performing JPEG transformation on the microcontroller module ESP32-CAM, which captures images with an embedded camera. The evaluation includes comparison of images compressed with JPEG alone and with the combination of the wavelet transform and JPEG, calculating the Peak Signal-to-Noise Ratio (PSNR) and Structural Similarity Index Measure (SSIM) coefficients.

Findings. A software module for implementing Haar wavelet transform on the ESP32-CAM microcontroller has been developed. The efficiency of using Haar wavelet transform as an additional method to JPEG for image compression and transmission is evaluated, considering the limited resources of the microcontroller. The size of compressed images and their quality was compared, assessing the size reduction of compressed images and the quality using PSNR and SSIM coefficients. The results showed that the additional use of Haar wavelet transform reduced the image size from 25% to times without significant loss in image quality, although the image processing time increased to 1 second.

Originality. A combined approach for image compression using multiple video processing methods, taking advantage of the capabilities of 32-bit microcontrollers with external memory, was proposed.

Practical value. The proposed image processing method can reduce the transmission time and energy consumption during transformations and image transmission, thereby extending the battery life of the sensor node. Moreover, this compression approach can maintain the quality of the original image with minimal data loss by reducing the amount of visual information lost during compression.