Research On Image Processing System Based On Air-conditioning Duct Cleaning Apparatus

Nowadays, the use of central air-conditioning has brought great convenience to people’s life, but because of the cleaning shortage to central air-conditioning ventilation duct, the pollution has seriously threatened to people’s health. It is an urgent requirement for effective cleaning to ducts, so the air-conditioning duct cleaning robot emerged.Machine vision has played a crucial role in cleaning robot system. In special environment of air conditioning ducts, if the robot vision is ineffective, workers will be likely to make wrong judgments and actions. So this paper focuses on the embedded image processing system of the machine vision. According to the performance requirements of the machine vision, denoising and compression encoding to the air-conditioning duct images are mainly researched, to ensure that images have a high quality, and can be transmitted or displayed in real time.Based on the theory of wavelet transform, the traditional wavelet threshold denoising algorithm and its shortcomings are analyzed. Then the improved method of the shrinkage function and new threshold selection method are proposed. The improved shrinkage function is a non-linear function based on the monotonic smooth log10(), which overcomes the intermittent and deviation of the traditional shrinkage function. The shrinkage threshold is choosing different threshold in different range of wavelet coefficients’correlation, so that the new threshold has a stronger self-adaptivity. EZW, SPIHT and SPECK are contrasted in the image compression encoding part, then SPECK are improved. After wavelet decomposition, the encoding sequence of S and I is exchanged to avoid the problem of cross calling in original SPECK, which recudes the complexity of the algorithm. Moreover,During scanning and sorting to block S, a List of Maximum Pixel(LMP) is introduced, which effectively solute the problem of rescanning, so the algorithm efficiency can be improved.Their effectiveness and superiority of above methods are verified by experiments. Improved denoising algorithm has stronger denoising ability and can retain more image details.And improved compression encoding one has a short running time and high compression ratio, and will obtain a higher quality image, which meet the requirements of the image transmission in this subject.